| 00:01.13 | TPABKA | hmm |
| 00:01.28 | *** part/#tomcat maxhugen (~maxhugen@CPE-144-132-240-161.nsw.bigpond.net.au) |
| 00:01.50 | TPABKA | seems like jk2 is so different that they even put workers2.properties into apache instead of tomcat... |
| 00:03.00 | TPABKA | bizarre :) |
| 00:03.10 | jasonb | jk != jk2 |
| 00:04.30 | TPABKA | hmm |
| 00:05.14 | TPABKA | except for the Listener directives i don't see a requirement to create Connectors for mod_jk |
| 00:05.17 | TPABKA | is that right? |
| 00:09.55 | jasonb | Rephrase your question, or be more verbose.. define "create". |
| 00:12.19 | TPABKA | well |
| 00:12.47 | TPABKA | do i need one of those Connecor sections in server.xml for mod_jk |
| 00:12.49 | TPABKA | ? |
| 00:14.01 | jasonb | Only if you want to connect to Tomcat in that way. :) |
| 00:15.21 | TPABKA | hell |
| 00:15.36 | TPABKA | some clear docs would be useful :) |
| 00:18.29 | jasonb | They sure would! |
| 00:40.05 | ceedo | you know the problem is that we are so bitter after figuring it out ourselves, no one wants to let anyone else in on the secret of how it works. |
| 00:45.11 | TPABKA | hmm |
| 00:45.26 | TPABKA | doesn't even seem like lotsa typing... |
| 00:45.27 | TPABKA | :) |
| 00:49.52 | ceedo | Writing documentation is an art. |
| 00:58.53 | Spline | bah |
| 00:59.01 | Spline | why am I always downloading specs in PDF? |
| 00:59.11 | Spline | You'd think sun would make all this stuff available in HTML like the API docx |
| 00:59.51 | kylev_ | ah, nothing warms the heart like deprecation warnings in brand-new code |
| 01:00.05 | jasonb | :) |
| 01:03.11 | kylev_ | ah, shit. |
| 01:03.19 | kylev_ | ok, I need to fire this guy |
| 01:03.37 | kylev_ | code with useful variable names like sql and sql1 |
| 01:06.35 | kylev_ | oh, it gets better |
| 01:06.40 | kylev_ | sql += .... |
| 01:07.33 | jasonb | boolean ever = true; |
| 01:07.40 | jasonb | for (; ever ; ) { |
| 01:07.53 | kylev_ | that's kind of a classic |
| 01:08.04 | kylev_ | at least it still reads like english |
| 01:08.06 | Spline | lol |
| 01:08.16 | kylev_ | I have a guy who always test for zero length strings |
| 01:08.22 | kylev_ | foo.length() < 1 |
| 01:08.42 | kylev_ | technically correct, but it always makes me stumble when reading his code |
| 01:08.51 | kylev_ | "Is he doing something more here?" |
| 01:08.54 | Spline | just in case its negative, eh? |
| 01:08.57 | jasonb | Does he check for null first? :) |
| 01:09.02 | kylev_ | no, just testing for a zero length string |
| 01:09.13 | Spline | yea I have problems with negative length strings all the time |
| 01:09.20 | jasonb | :) |
| 01:10.17 | Spline | Heh, I know! Lets check if sql.length() < 1, then catch the NullPointerException! |
| 01:15.32 | kylev_ | Fuck that, man. This guy never catches anything other than Exception |
| 01:15.49 | kylev_ | if he could enclose entire classes in one try/catch, he would |
| 01:27.19 | jkk | ceedo: you there? |
| 01:39.25 | Spline | oh fsck |
| 01:39.33 | Spline | I got all the way through migrating this ^%&*^)&(& app |
| 01:39.46 | Spline | and now I discover they are using Websphere proprietary classes in their JSP |
| 01:40.05 | Spline | Somebody smurf my system |
| 01:55.17 | jasonb | So much for servlet container independence when someone does stuff like that.. |
| 01:56.31 | Spline | yea |
| 01:56.39 | Spline | they didnt even isolate the non-portable code |
| 01:56.46 | Spline | just tossed into a few JSPs |
| 01:59.42 | Spline | hey, do sessions persist across webapps on a single tomcat instance? |
| 02:11.20 | jasonb | No. Each webapp has its own sessions, except for when you use the SingleSignOnValve. |
| 02:12.06 | Spline | ah yes |
| 02:12.12 | Spline | thx |
| 02:13.25 | jasonb | Yup. |
| 03:47.11 | *** join/#tomcat Orfindel (Orfindel@ACA63268.ipt.aol.com) |
| 03:47.16 | Orfindel | I have perl script that reads a table (based on some parameters), and dynamically produces a result in an HTML format. The only problem I have is that if the user refreshes the page, it does the same thing again. Is there a way to clear the URL address, or change it to something else. Not necessairly to execute, but if the user refreshes the page, it will go there and not envoke the perl script again. ideas? |
| 04:29.45 | *** join/#tomcat jasonb (~jasonb@adsl-66-124-73-250.dsl.sntc01.pacbell.net) |
| 05:01.02 | *** join/#tomcat jrandom (nobody@cs6669249-49.austin.rr.com) |
| 05:02.04 | *** join/#tomcat jasonb_ (~jasonb@adsl-66-124-73-250.dsl.sntc01.pacbell.net) |
| 05:02.35 | jrandom | Is this the best place to ask questions about JSTL and other taglibs, or should I go somewhere else? |
| 05:12.57 | *** join/#tomcat IOSninja (~ninja@OL10-33.fibertel.com.ar) |
| 05:13.09 | IOSninja | Hey There |
| 05:13.19 | IOSninja | Is this a Tomcat help channel or something? |
| 05:37.23 | *** join/#tomcat jkk (www@dyn-mcb-168-57.dyn.columbia.edu) |
| 05:38.19 | jkk | re all |
| 05:43.45 | jasonb | reeeee |
| 05:44.17 | IOSninja | hey |
| 05:52.31 | IOSninja | does anyone know how to set the maximum number of processed that the ajp13 connector can spawn |
| 05:54.25 | jkk | IOSninja: most likely an attribute from conf/server.xml |
| 05:55.22 | IOSninja | well, I played with all that stuff |
| 05:55.26 | IOSninja | nothing seems to work |
| 05:55.44 | jkk | read the docs |
| 05:55.46 | IOSninja | if I set the Maximum Processes to 100, it spawns three times more |
| 05:55.54 | jkk | it should be as simple as maxConnectors |
| 05:56.02 | jkk | unless you're not setting the right attribute |
| 06:06.28 | IOSninja | <PROTECTED> |
| 06:06.28 | IOSninja | <PROTECTED> |
| 06:06.28 | IOSninja | <PROTECTED> |
| 06:06.34 | IOSninja | maxProcessors |
| 06:06.48 | IOSninja | I set it to 230, but it's the same than setting it to 20 |
| 06:06.58 | IOSninja | it starts spawining till it explodes |
| 06:08.45 | IOSninja | I mean, it's totally insane... |
| 06:09.15 | jkk | i dunno man |
| 06:09.16 | IOSninja | If do lots of reloads in the browser, the site goes down |
| 06:09.40 | IOSninja | I've put mod_dosevasive and still |
| 06:10.20 | IOSninja | I've even twaked the /proc/sys/net/ipv4 params to make timeouts shorter |
| 06:10.22 | IOSninja | and nothing worked |
| 08:43.22 | *** join/#tomcat Binbo (~Binbo@p157.nas3.akl.callplus.net.nz) |
| 08:45.17 | Binbo | who has used dyndns service to make tomcat available to net? |
| 08:46.48 | Binbo | i have a domain name available, which file do i configure to make tomcat host that name? |
| 09:16.47 | *** join/#tomcat Remm_ (~chatzilla@AToulouse-206-1-14-247.abo.wanadoo.fr) |
| 09:25.37 | Binbo | meow |
| 10:39.47 | *** join/#tomcat jkk (www@dyn-mcb-168-57.dyn.columbia.edu) |
| 11:43.39 | *** join/#tomcat jkk (www@dyn-mcb-168-57.dyn.columbia.edu) |
| 13:22.11 | *** join/#tomcat BaD_SaNta (BaD_SaNta@212088094106.sonofon.dk) |
| 13:43.30 | *** join/#tomcat Freix (~robbie@host191-6.pool80180.interbusiness.it) |
| 15:10.39 | *** join/#tomcat TheDraw (~Lyss@h4xed.j00.and.st0le.all.yourpr0n.com) [NETSPLIT VICTIM] |
| 15:23.15 | *** join/#tomcat jasonb (~jasonb@adsl-66-124-73-250.dsl.sntc01.pacbell.net) |
| 15:24.15 | *** join/#tomcat xerov (123@ip68-4-0-92.pv.oc.cox.net) |
| 15:24.45 | xerov | im getting status 404 for a servlet i know is there, any way i can debug the problem? |
| 15:24.55 | xerov | i dont see anything unusual in the logs either |
| 15:25.38 | xerov | its a regular turbine servlet, works on development machine |
| 16:27.05 | *** join/#tomcat bugfixer (~bugfixer@42gis175.gulftel.com) |
| 16:59.33 | *** join/#tomcat PW_Wrk_Away (~Pointwood@fw.agropark.dk) [NETSPLIT VICTIM] |
| 17:03.50 | *** join/#tomcat jasonb_ (~jasonb@adsl-66-124-73-250.dsl.sntc01.pacbell.net) |
| 17:40.34 | *** join/#tomcat jkk (www@dyn-mcb-168-57.dyn.columbia.edu) |
| 17:46.37 | *** join/#tomcat mrb__ (~MRB@216.54.81.98) |
| 17:59.28 | *** join/#tomcat jkk (www@dyn-mcb-168-57.dyn.columbia.edu) |
| 17:59.39 | jkk | ceedo: you there? |
| 18:33.36 | ceedo | yes |
| 19:01.50 | hellabot | "destr0" pasted Pear DB Access Objects at http://www.hellaweb.com:8888/44 |
| 19:52.18 | *** join/#tomcat jasonb (~jasonb@nat1.sp.collab.net) |
| 19:53.50 | ceedo | poor jkk |
| 20:01.19 | *** join/#tomcat jkk (www@dynamic-156-91.dyn.columbia.edu) |
| 20:02.00 | jkk | ceedo: awake? |
| 20:02.28 | ceedo | Yeah |
| 20:02.57 | jkk | ceedo: i was wondering if you wouldn't mind sending me your server.xml where you define your custom connector? |
| 20:03.11 | ceedo | sure, what is your email address. |
| 20:04.12 | ceedo | mm I like dark wet nooks. |
| 20:04.34 | ceedo | ok should be in your box |
| 20:04.55 | jkk | thanks! |
| 20:05.22 | *** join/#tomcat Mickael_Tomcat (~chatzilla@sky.planb-inc.com) |
| 20:05.34 | ceedo | Are you having problems getting it to load your connector? |
| 20:08.18 | jkk | ceedo: yeah |
| 20:08.29 | jkk | ceedo: i'm gonna try for the embedded approach |
| 20:09.12 | ceedo | What web server are you trying to get this to work with? |
| 20:09.26 | ceedo | embedding the jvm incidently is very complicated and nonobvious |
| 20:09.36 | ceedo | JNI is simple but you have to do some cute things to make it work with catalina |
| 20:21.07 | jkk | eh |
| 20:21.13 | jkk | uh.. i'm embedding tomcat i think |
| 20:21.17 | jkk | not the entire jvm |
| 20:21.58 | ceedo | oh |
| 20:22.01 | ceedo | I am not sure I follow the distinction |
| 20:22.16 | ceedo | oh are you using a java web server? |
| 20:22.29 | ceedo | and you just want both to run in the same jvm? |
| 20:22.55 | jkk | well kinda.. i have my own non-http protocol and i want to write servlets for it |
| 20:23.24 | jkk | so I'm hoping i can write my own Connector, my own Host and my own Servlet extension and get all the wonderful class loading, security, and logging stuff of tomcat |
| 20:24.19 | ceedo | ohh |
| 20:24.24 | ceedo | you don't even need to write host |
| 20:24.32 | ceedo | All you need is to write a connector and those two request/response objects. |
| 20:24.55 | ceedo | What is the benefit of your protocol? |
| 20:25.07 | jkk | ceedo: it's a broadcast form of http |
| 20:25.21 | jkk | ceedo: well i need my own Host to do some basic 'transaction' stuff |
| 20:25.54 | ceedo | Everything for accessing the protocol data is in the request/response objects. |
| 20:26.39 | ceedo | What do you use it for? |
| 20:26.54 | jkk | ceedo: each request to a URL gets sent to many different HTTP-like resources. One of those resources is the 'real' resource and others are just mirrors. when the real resource responds, the mirrors process 'request/response' to update their state to reflect the state of real resource |
| 20:27.08 | jkk | ceedo: well who knows |
| 20:27.35 | ceedo | that is an odd mechanism. |
| 20:27.48 | jkk | ceedo: odd? |
| 20:28.08 | ceedo | An odd design |
| 20:28.46 | jkk | it's quite inefficient |
| 20:29.00 | ceedo | Well a number of issues jump to mind |
| 20:29.09 | jkk | ceedo: like? |
| 20:30.02 | ceedo | well I am presuming the mirror is supposed to be able to respond of the real resource is down with it's local copy, but, that's inefficient in the time domain and requires the implementation of a collision protocol |
| 20:30.33 | ceedo | plus I presume it's actual real broadcast networking which is of course inefficient |
| 20:31.00 | ceedo | but I clearly have no idea what it's really supposed to do |
| 20:32.11 | jkk | ceedo: it's based on a real pub/sub event network which is actually really efficient. we can do 80,000 events/sec. We're still figuring out the best way to do a decentralized collision protocol, you're right. right now all the mirrors get together and vote on who gets to be the real resource but this has many transaction/timeout issues. |
| 20:34.14 | ceedo | Why was it not possible to make all the nodes authoritative through a backing store of some kind? |
| 20:35.37 | jkk | ceedo: not sure what you mean... though we don't want a centralized solution. the idea is even though you have massive dependencies and coordination, the network itself should be just like the HTTP/Web, that is scale free. A resource shouldn't care how many mirrors it has. |
| 20:37.37 | jkk | of course it's not scale free, there are real problems with the underlying event protocol where you have one client sending a request to potentially thousands of destinations |
| 20:37.56 | ceedo | I think what I am trying to say is that I would have tried to manage the availabiltiy issue through the design of the storage architecture |
| 20:38.06 | ceedo | so I am curious how your situation came to pass. |
| 20:38.09 | jkk | ceedo: storage architecture? |
| 20:38.33 | jkk | ceedo: my situation? oh, this isn't for a commercial product.. it's research.. programming systems lab @ columbia university |
| 20:39.14 | ceedo | Well here I have let's say 35 nodes that respond to network requests and provide access to a resource |
| 20:39.20 | ceedo | let's call them webservers |
| 20:39.49 | ceedo | If the resource is not located on the processing nodes (feps) but rather in a backing store than individually they are not important to the availability on the platform |
| 20:40.04 | jkk | yeah the are essentially webservers. don't call them nodes, call them 'resources'. they're just like HTTP URL-addressable objects |
| 20:40.36 | jkk | ceedo: oh i see what you're getting at |
| 20:40.51 | *** join/#tomcat Esaj (~jason@jdd30.joh.cam.ac.uk) |
| 20:40.52 | ceedo | networking is always the highest overhead in a real-world design, so a single storage platform can handle many times more webservers or other application processors |
| 20:41.11 | jkk | ceedo: well yes and no |
| 20:41.19 | ceedo | for example I do several hundred million web requests a day and I spend more time in kernel dealing with the packet overhead than in the application |
| 20:41.21 | Esaj | Is it possible to set a password for SMTP authentication in the server.xml file? |
| 20:41.29 | ceedo | on the next tier (the backing store) the machines idle all day |
| 20:41.44 | jkk | ceedo: the problem we are concerned with is not so much high availability but massively distributed collaboration & coordination. |
| 20:42.02 | ceedo | If the hardware has to be within the same multicast domain that is hardly distributed, though. |
| 20:42.24 | Esaj | Or is it possible to make an Authenticator object available, using server.xml? |
| 20:42.25 | ceedo | but I thought that you had a commercial application that you were trying to solve for scalability and availability |
| 20:42.33 | jkk | ceedo: that is, distributed state. high availbility of a single machine doesn't matter at all, if all your information is replicated across tons of different machines. |
| 20:42.59 | ceedo | yes but I question the price/performance of that archiecture vs. one with a clustered backing store |
| 20:43.19 | ceedo | thankfully though you don't have a comptroller to worry about when writing papers ;) |
| 20:43.23 | jkk | ceedo: in fact, part of the idea is.. if you're in the military in a war and the command center gets blown up.. it doesn't matter how many clusters you have |
| 20:43.33 | jkk | they are all dust now |
| 20:43.51 | ceedo | that is a really great architecture if nodes have infinite storage |
| 20:44.07 | ceedo | and also if you say that the network will still function in disparate groups even if your backing stores are gone |
| 20:44.18 | jkk | ceedo: why do you say a node needs infinite storage? |
| 20:44.52 | ceedo | Well the viability of that architecture after the disaster will depend on the size of the local storage that's left in the resulting domain |
| 20:45.18 | ceedo | which in turn will probably have to do with the types of queries that were being made prior which were refreshing the mirror data |
| 20:45.18 | ceedo | afk |
| 20:49.01 | jkk | ceedo: actually that is a good point about infinite storage i hadn't thot of... it is possible of course to have very long txs... |
| 20:51.59 | ceedo | sorry someone walked into my office to seek my advice on garage door openers |
| 20:54.10 | ceedo | in cache architectures in general there is usually the problem of some consumer coming along sooner or later and requesting a lot of data which causes a lot of data get get mirrored, driving other useful data out of the cache |
| 20:54.37 | ceedo | for a random practical example this is what dba's struggle with every day trying to tune oracle to work for both transactional and batch queries |
| 20:54.45 | ceedo | so your system also needs to be somwhat smart |
| 20:55.00 | ceedo | like if the solider in the trenches is looking at a lot of porn it shouldn't through out the iraqi troop movments |
| 20:55.08 | ceedo | because porn won't do them any good when the bombs come and break the communications grid |
| 20:55.22 | ceedo | that is where infinite storage would come into play ;) |
| 20:55.43 | ceedo | er throw out |
| 20:57.29 | ceedo | If you can solve this problem, patent it |
| 20:57.44 | ceedo | because everyone who has made a cluster from nema to oracle to filesystem replication wants the answer |
| 20:59.30 | jkk | ceedo: heh |
| 21:00.04 | jkk | ceedo: well the trivial solution is not to demand infinite storage from any single node but to demand 'near infinite' mirroring |
| 21:00.32 | ceedo | sure infinite nodes with finite storage ;) |
| 21:00.41 | jkk | exactly :) |
| 21:01.07 | jkk | storage doesn't scale but near-infinite replication is a very real possibility |
| 21:01.49 | ceedo | with near-infinite bandwidth |
| 21:01.59 | jkk | yes bandwidth is a very real problem |
| 21:03.17 | jkk | but not one i really care about :) |
| 21:03.47 | ceedo | If you are willing to have infinite anything then you will be fine. |
| 21:04.31 | ceedo | no more money for me, I'm giving it up. |
| 21:05.43 | jkk | ceedo: well the the question is not really infinite |
| 21:06.14 | jkk | obviously you can't inifnite anything. the question is what is 'scale-free'... in other words, what do you allow to approach infinity |
| 21:06.34 | jkk | you can't ask a single computer to have infinite computational power or storage... |
| 21:06.37 | ceedo | well that's an indusry question. |
| 21:06.47 | jkk | but you can ask that you have infinite computers of finite power and storage |
| 21:06.50 | ceedo | and right now storage is increasing more than either processing power or bandwidth |
| 21:06.57 | ceedo | bandwidth is in fact increasing the least of the three by far I think. |
| 21:07.00 | jkk | because there is no 'hard limit' on the number of computers |
| 21:07.17 | ceedo | well if you can have 2 computers with n then you can have 1 computer with n2 space |
| 21:07.48 | jkk | ceedo: yes and bandwidth will always increase and rise to the occassion...we hope |
| 21:08.44 | ceedo | well all three facets are rising, I am saying though that communications, particularly on the wan, is not keeping up. |
| 21:08.49 | ceedo | Networking is not doubling like disk or cpu. |
| 21:08.57 | ceedo | moore's law seems to not apply to wan's or even lans |
| 21:09.01 | jkk | ceedo: well perhaps on the global internet... fiber and gigabit ethernet can solve many bandwith problems which is what makes clustering possible |
| 21:09.39 | jkk | ceedo: bandwidth doesn't need to.. you can always just lay more fiber which is the point |
| 21:10.09 | ceedo | you can always buy more drives, or more computers, too by that logic |
| 21:10.22 | ceedo | My point is simply that from a cost/value standpoint disk is increasing the fastest |
| 21:11.03 | jkk | yes exactly.. you can always buy more computers. that is why i say the number of computers within the network is scale-free .. just like the web. any given computer doesn't have to have infinite computational power or memory like in centralized solutions. |
| 21:11.45 | ceedo | or you could buy more disk for n computers |
| 21:11.55 | ceedo | by that logic 1 computer is scale-free |
| 21:12.10 | ceedo | ot |
| 21:12.28 | ceedo | it's ok though once they get 1.8" 180gb drives in a couple years your soliders will have plenty of room for all the porn mirrors |
| 21:12.29 | *** join/#tomcat {aaron} (blah@aeolus.cit.cornell.edu) |
| 21:12.29 | jkk | ceedo: yes i think you are right. networked storage solutions are very cost effective and of course that is what many corps are doing, just creating these big SANs but this is not a solution to the problem of distributed state.. it is highly centralized. |
| 21:12.47 | ceedo | there are decentralized solutions for that architecture as well. |
| 21:13.02 | ceedo | the point of using a san network is force multiplication by offloading the application overhead on a n+1 tier |
| 21:13.27 | ceedo | but yes that is not battlefield survivable |
| 21:13.32 | {aaron} | anybody familiar with the webapp or jk connectors? |
| 21:13.48 | {aaron} | can these be used off the shelf to create a custom java request server? |
| 21:15.03 | jkk | {aaron}: http://jakarta.apache.org/tomcat/tomcat-4.1-doc/config/ |
| 21:15.14 | jkk | {aaron}: most likely yes.. |
| 21:16.52 | jkk | ceedo: but you bring a good point, i think.. and in response i think not being cost effective or "efficient" is the sign of a successful network system. for example, you could hardly call HTTP "efficient" but look how well it has scaled |
| 21:17.36 | jkk | when you focus on price/performance you end up making compromises which in turn limit the flexibility of your solution |
| 21:17.58 | jkk | that is why i have no problem pretending that there is such a thing as infinite bandwidth and infinte computers |
| 21:18.36 | ceedo | http is very efficient. |
| 21:18.48 | {aaron} | eh? |
| 21:19.01 | ceedo | Don't confuse elegance for efficiency ;) |
| 21:19.41 | {aaron} | man, webapp connector doesn't even use a thread pool... |
| 21:20.07 | jkk | http is notoriously inefficient and it is remarkably bad for real world business scenarios.. not even any support for transactions.. but that's ok |
| 21:20.10 | jkk | 'less is more' |
| 21:20.36 | ceedo | inefficient in what sense? |
| 21:21.23 | ceedo | And I think that is two seperate topics; http addresses the resource distribution that it was designed to do, that it doesn't work for transactional applications is an artifact of people trying to use it for something it wasn't intended to be used for. |
| 21:21.42 | ceedo | That doesn't speak for or against the protocol or it's applicability but rather just the tenacity of people to force a square peg into a round hole |
| 21:21.51 | {aaron} | ahh...all i have to do to use warp connector is create my own Container impl...nice :) |
| 21:21.55 | jkk | well i do not have the time to get too into it, but it is inefficient in the sense that there are many more faster, more efficient, more robust ways to do what HTTP does and allow for things like guaranteed delivery and fast failure. |
| 21:22.07 | jkk | {aaron}: warp connector? |
| 21:22.20 | ceedo | yes, neither of which were problems it was designed to address. |
| 21:22.20 | {aaron} | yes, the webapp connector |
| 21:22.34 | jkk | {aaron}: what protocol does warp implement? |
| 21:22.40 | {aaron} | um, warp? heh |
| 21:22.45 | jkk | ohhh |
| 21:22.46 | jkk | heh |
| 21:22.54 | {aaron} | it's yet-another-flavor-of-JK |
| 21:23.15 | {aaron} | i dont' care *how* it works, as long as I can get apache to talk to it, and hence my custom-built implementation |
| 21:23.16 | jasonb | It's technically a different protocol from JK, but the theory is based on JK. |
| 21:23.19 | {aaron} | yes |
| 21:23.34 | ceedo | jkk I think the only thing we have established by using http as a case study is that people will do what is cheapest ;) |
| 21:23.41 | {aaron} | i'm not sure whata apache wants to do with it, the docs say: "read more about warp here" and then the link is dead :/ |
| 21:23.44 | jasonb | Actually, not based on JK, but JServ. :) |
| 21:23.46 | ceedo | and has the least friction |
| 21:23.53 | jkk | ceedo: yes but of course, no computer system will satisfy everyone everywhere.. but we would like to have as many customers as possible. but yes, you are right, it is not totally fair to call http 'inefficient' in that it had very modest goals to begin with |
| 21:24.15 | ceedo | yeah http is also inefficient in, say, brewing coffee |
| 21:24.17 | ceedo | :) |
| 21:24.28 | {aaron} | i'm sure berners lee didn't invision the amount of crud we ram over HTTP these days |
| 21:24.39 | ceedo | I don't think berers lee had a clue at all. |
| 21:24.42 | ceedo | er berners |
| 21:24.45 | {aaron} | heh |
| 21:25.03 | ceedo | It's only by a fluke of development that we don't all use gopher right now for our multimedia content |
| 21:25.16 | jkk | ceedo: no i don't think it's a fluke |
| 21:25.23 | jkk | the gopher model was very silly |
| 21:25.28 | ceedo | the http model is very silly |
| 21:25.47 | jkk | the combination of URLs and a finite set of verbs (GET,PUT,POST,DELETE, etc..) is very powerful |
| 21:25.51 | ceedo | but the http model gave people porn faster and so succeeded because the gopher people were lagging behind on delivering that, although they did manage to do it too late. |
| 21:26.08 | ceedo | except, no one uses anything but get and post and sometimes head. :) |
| 21:26.10 | {aaron} | gopher has porn?? |
| 21:26.16 | ceedo | aaron no that's why you use http |
| 21:26.29 | {aaron} | you guys are so 90s...that's what P2P is for |
| 21:26.33 | {aaron} | get with it |
| 21:26.37 | ceedo | It's my theory that any technology develops towards the point of delivering porn |
| 21:26.46 | jkk | ceedo: that is not true... many systems exist which take advantage of the full range of HTTP. |
| 21:26.52 | {aaron} | so then that porn plugin should be coming any day for emacs |
| 21:27.00 | ceedo | jkk, name one |
| 21:27.05 | {aaron} | webdav |
| 21:27.06 | jkk | {aaron}: it already exists, emacs has a bult in browser |
| 21:27.07 | ceedo | one that is in commercial use somewhere |
| 21:27.35 | {aaron} | i'm no fan of http |
| 21:27.45 | jkk | ceedo: well i could give you names but you would recognize few. webdav is one, SIMP is another you might know of |
| 21:27.57 | {aaron} | or rather i'm no fan of using http for interactive applications that are way more complicated than document delivery |
| 21:28.16 | {aaron} | "document" is the wrong metaphor for many things |
| 21:28.30 | jkk | ceedo: but the point is, it is in fact possible through a combination of the right resources and the basic verbs (GET,PUT,POST,DELETE) to model ANY process or network interaction you imagine |
| 21:28.39 | ceedo | jkk of course you can |
| 21:28.43 | ceedo | http is better than gopher |
| 21:29.00 | ceedo | but, we would still be using gopher now if not for the fact they were too late developing graphical clients. |
| 21:29.14 | {aaron} | jkk: yes, but it's still a better idea to create new protocols that are more tailored to their use |
| 21:29.17 | ceedo | I think it's clear that the marketplace doesn't select technology based on anything other than price/performance which is my only argument. |
| 21:29.35 | {aaron} | wait no |
| 21:29.41 | {aaron} | the marketplace is full of stupid people |
| 21:29.52 | ceedo | The marketplace is full of stupid people that drive big cars. |
| 21:29.53 | {aaron} | i'd say inertia is stronger than any other force alone |
| 21:30.00 | {aaron} | people use what already works |
| 21:30.02 | jkk | {aaron}: actually it's not. new protocols like SOAP and the like are very bad from a 'top-down' point of view. |
| 21:30.10 | ceedo | Yeah but why does inertia carry weight? Because what is already built and deployed is cheaper ;) |
| 21:30.12 | jkk | ceedo: well that is why we do not let the marketplace decide |
| 21:30.30 | ceedo | jkk I am part of the marketplace |
| 21:30.32 | ceedo | so |
| 21:30.33 | {aaron} | yeah, SOAP is pretty bad |
| 21:30.39 | {aaron} | it's still over HTTP which makes it even worse |
| 21:31.11 | jkk | ceedo: all the new networking technologies begin in research labs where very smart people waste away their days thinking about this :) |
| 21:31.25 | jkk | of course, you still get things like SOAP |
| 21:31.37 | ceedo | jkk to believe that I would also have to believe that nothing better than http existed in a lab since 92 |
| 21:31.46 | ceedo | and I refuse to believe you people in acadamia are that dim |
| 21:31.51 | jkk | because every very smart people can be very stupid and repeat all the same mistakes over again |
| 21:32.19 | jkk | ceedo: i keep telling you, you underestimate http. there are many messaging protocols that you might call 'better' than http |
| 21:32.32 | jkk | but could they support a system the size of the internet? |
| 21:32.33 | ceedo | I underestimate http? I was the one estolling it's virtues |
| 21:32.46 | {aaron} | yeah, i'm not even sure who was on what side of the argument any longer |
| 21:32.55 | ceedo | http has nothing to do with the scale of the internet |
| 21:33.03 | {aaron} | http should be left to serving documents |
| 21:33.12 | jkk | i am not arguing, i was just saying ceedo's focus on price/performance is misguided |
| 21:33.22 | jasonb | How much ram does this internet thing have? :) |
| 21:33.36 | {aaron} | jasonb: apparently not enough, sometimes it is sooo slow |
| 21:33.36 | ceedo | jason jkk hopes a near-infinite amount ;) |
| 21:33.38 | jkk | ceedo: heh, i think http is the deciding factor of the size of the infinite |
| 21:33.51 | ceedo | jkk hardly |
| 21:33.55 | ceedo | ip clearly is |
| 21:34.03 | {aaron} | ipv6 |
| 21:34.21 | jkk | letting any computer get/put data from any other computer using URLs and a simple set of verbs like i said, is a very powerful mechanism |
| 21:34.34 | jkk | ip is an implementation detail :) |
| 21:34.38 | ceedo | sure but it's not what enables the internet to have billions of machines |
| 21:34.57 | jkk | it does not matter how data is actually moved around and nodes are addressed has little to do with the nature of the network |
| 21:35.15 | ceedo | there are other protocols that are better and others that are worse but no one cares which one is actually used and it has little impact |
| 21:35.38 | ceedo | on the other hand the total capacity of the internet addressing scheme has a lot of impact :) |
| 21:35.48 | ceedo | as do the routing protocols |
| 21:35.53 | jkk | ceedo: you see, there are in fact two ways of thinking about what a computer network is. you might say a computer network is a bunch of computers connected and moving around data between them. but this is the simple, 'bottom-up' view |
| 21:35.57 | ceedo | BGP has more to do with the internet being big than http ever would |
| 21:36.04 | jkk | ceedo: you can't actually describe the network like that |
| 21:36.31 | jkk | ceedo: no.. it's very easy to always change the addressing scheme to allow for more addresses without changing the network |
| 21:36.38 | ceedo | hahahaha |
| 21:36.38 | ceedo | ok |
| 21:36.45 | ceedo | that's why we're switching to ipv6 next tuesday? :) |
| 21:37.07 | jkk | ceedo: yes and that's why it doesn't matter to most other systems that you are switching |
| 21:37.22 | ceedo | look if you want to thank anything for the internet being vast |
| 21:37.26 | ceedo | thank bgp, ospf, and rip |
| 21:37.31 | ceedo | but you just call those "details" |
| 21:37.35 | ceedo | I just call http a "detail" |
| 21:37.38 | jkk | ceedo: like i said, how the network actually works is irrevelant |
| 21:37.52 | ceedo | how the computers phrase a request and get an answer is irrelevant too |
| 21:37.53 | ceedo | as long as they can |
| 21:37.55 | jkk | ceedo: well i think we are using the word differently |
| 21:38.29 | jkk | to me, a computer network is a function that describes how to take a distributed set of turing machines from one state to the next |
| 21:38.41 | jkk | to you, a computer network is just a way of shuffling bits around |
| 21:38.51 | ceedo | no, that's not true. |
| 21:38.59 | jkk | so you might focus on ip, but i focus on the 'model' ie http |
| 21:39.04 | ceedo | because you don't agree with me, don't simplify me. |
| 21:39.25 | jkk | im not, im simply explaining what i mean when i call ip a 'detail' of the network |
| 21:39.31 | ceedo | http isn't a model, client server request/response is a model |
| 21:39.37 | ceedo | http is a detail :p |
| 21:39.52 | jkk | http is the model that defines the request/response :) |
| 21:40.01 | ceedo | req/res existed long before http dude. |
| 21:40.19 | jkk | yes of course |
| 21:40.32 | jkk | but there is a very real mathematical difference between a SOAP request and an HTTP request |
| 21:40.46 | ceedo | yes and which one is used is just as much an irrelevant detail as ip |
| 21:40.59 | jkk | and there is a very real mathematical reason for why you can put a SOAP request inside an HTTP request but you can't really do the reverse |
| 21:41.13 | jkk | ceedo: well like i said, it depends on what the network really is |
| 21:41.22 | ceedo | why do you keep bringing the network into this? |
| 21:41.51 | ceedo | I feel some need to restate my point: the choice of http was abitrary and driven by market forces |
| 21:41.53 | jkk | ceedo: because that is what it's all about. before you can argue about the merits of protocols you must be able to answer the question 'what is a computer network?' |
| 21:41.59 | ceedo | it was just the best of breed at that moment |
| 21:42.10 | ceedo | now it has all the interia it needs to last for years |
| 21:42.13 | {aaron} | or maybe "only of breed" |
| 21:42.39 | ceedo | jkk you are just going around in circles here. |
| 21:42.58 | jkk | ceedo: perhaps, we shall have to agree to disagree :) i think the triumph of http wasn't inevitable but happened due to the technical merits of http over all other protocols |
| 21:43.09 | {aaron} | http is good because it imposes so few semantics, but http is also bad *because it imposes so few semantics* |
| 21:44.57 | jkk | ceedo: and in fact there have been many cases where people have said 'we can do better than http, let's come up with a domain-specific solution that is customized for the problem we're trying to solve' eg WAP and ROCK |
| 21:45.24 | ceedo | I think that's an argument that http was just there at the right time and sort of did the right thing. |
| 21:45.25 | jkk | ceedo: but in the end, people always come back to http.. of course that may be due to http's inertia like you say, but i dont think that's just it |
| 21:45.44 | {aaron} | i think that is it |
| 21:45.45 | ceedo | jkk, you think it's choice and not the 9438249028349238 copies of internet explorer? |
| 21:46.11 | {aaron} | yes, the ubiquitous client really drives it |
| 21:46.22 | {aaron} | show me a ubiquitous client for anything and I will show you a lasting protocol |
| 21:46.27 | jkk | ceedo: well that certainly doesn't hurt the case for http |
| 21:46.32 | {aaron} | of course the situation is a catch22 |
| 21:46.34 | ceedo | aaron, and the protocol can even be totally ass :) |
| 21:46.59 | jkk | ceedo: but you cannot really explain away http's success like that, saying that 'X is successful because X is successful' is a tautology not an explanation |
| 21:47.14 | ceedo | X is succesful because it was the only option at the time. |
| 21:47.18 | ceedo | I can very well say that. |
| 21:47.31 | ceedo | And then I can also very well say that it has now become entrenched in the market place as a result of early penetration. |
| 21:47.38 | jkk | you can, but that is not a very good explanation and will not lead to any interesting papers :) |
| 21:47.39 | ceedo | "first to market" is biz 101 :) |
| 21:47.42 | ceedo | no |
| 21:47.45 | ceedo | but it will lead to money |
| 21:47.50 | ceedo | I make money not write papers. |
| 21:48.01 | ceedo | This is always the critical difference between acadmia and business and where it all falls apart. |
| 21:48.03 | jkk | ceedo: but see that is the point, HTTP was _not_ first to market |
| 21:48.25 | ceedo | It was the first to market with a graphical client that could embed images along with style information |
| 21:48.25 | {aaron} | it was about the same time the PC revolution was exploding |
| 21:48.30 | {aaron} | or rather the internet |
| 21:48.31 | {aaron} | so yeah |
| 21:48.38 | jkk | in fact like you said earlier, when HTTP showed up there were existing protocols that in fact did 'more' than HTTP |
| 21:48.43 | ceedo | If you know of something else that predates it that also could do that across platforms, I would like to hear it.. |
| 21:48.49 | ceedo | gopher could do rich types but not on a single page |
| 21:48.56 | ceedo | not until after http. |
| 21:49.33 | {aaron} | "viewing documents" was what made the internet useful to average people, average people drive the market |
| 21:49.41 | {aaron} | so that's why we have it |
| 21:49.54 | ceedo | aaron he's googling to try to figure out what predates http that meets my criteria ;) |
| 21:49.56 | ceedo | give him a minute |
| 21:50.15 | jkk | ceedo: but ultimately, almost any 'end-to-end' protocol can simulate any other 'end-to-end' protocol. like you said, once you have a client and a server you pretty much have everything you need. it's a question of the model a protocol presents |
| 21:50.23 | {aaron} | it doesn't matter even if there was one, because once users got a hold of a useful application (the "web") they took off with it |
| 21:50.51 | jkk | {aaron}: but why did the web need http? why not anything before? |
| 21:50.52 | ceedo | aaron well he could damage my argument if he could prove there was a comperable technology and so http was "selected" via it's technical merits |
| 21:51.04 | {aaron} | jkk: ask Tim Berners Lee |
| 21:51.18 | {aaron} | jkk: i have no idea why the "web" was done over HTTP rather than FTP |
| 21:51.28 | {aaron} | or any other protocol |
| 21:51.31 | jkk | {aaron}: you see, there is a very real reason that you can look at HTTP and imagine something called a 'browser' that just roams around, talking to different resources. it is because of HTTP's focus on URLs and it's finite verbs. |
| 21:51.53 | ceedo | smtp is verb based too |
| 21:51.55 | {aaron} | agreed |
| 21:51.57 | ceedo | but you don't see them using that |
| 21:52.06 | jkk | {aaron}: but for many other protocols, the concept of a 'browser' is totally alien. the client is expected to share an object model with the server like in most distributed object/RPC protocols |
| 21:52.09 | ceedo | I think it's obvious that lee thought he was on to something novel |
| 21:52.13 | {aaron} | well, HTTP is certainly suited for document delivery, that's what it was designed for |
| 21:52.18 | ceedo | but that doesn't change the fact the market selected it |
| 21:52.20 | jkk | ceedo: yes, synchrocity is another thing |
| 21:52.25 | ceedo | We've all written protocols in here |
| 21:52.29 | {aaron} | jkk: *nod* |
| 21:52.31 | ceedo | raise your hand if one of them is used by the entire world |
| 21:52.58 | {aaron} | if you only really have one choice, i'm not sure that is exactly "selection" |
| 21:53.11 | ceedo | aaron precisely my point. |
| 21:53.30 | ceedo | but they did select it in the sense that they didn't hire jkk to develop something else better |
| 21:53.32 | jkk | ceedo: the market didn't select http, the market was desperate for a protocol that would allow the exchange of a massive amounts of information in a highly structured but loosely coupled manner that was inherently platform dependent. |
| 21:53.56 | ceedo | jkk, so then you admit that http's success comes from the marketplace |
| 21:54.08 | ceedo | You are on the verge of agreeing with me now |
| 21:54.13 | {aaron} | truthfully, I don't think anybody had any rational reasons for choosing any specific technology early on...everybody was just dazzled by "Information Superhighway" this, and "Internet Age" that.. |
| 21:54.16 | ceedo | you can do it just step over |
| 21:54.26 | ceedo | aaron that wasn't until 94 when AOL got online |
| 21:54.37 | jkk | ceedo: i think you are agreeing with me. http made the browser possible and what the market _needed_ was the browser. there was no 'choice' or 'selection' going on. |
| 21:54.51 | {aaron} | ceedo: and i think that greatly drove the acceptance of HTTP and HTML... etc. |
| 21:55.13 | jkk | ceedo: and in fact, you see, that is the point. if you think a protocol is just a way of moving bits around you cannot explain the success of http because there were many protocols that moved bits around before and after http |
| 21:55.42 | ceedo | jkk, now you are 180ing |
| 21:56.11 | jkk | ceedo: i am repeating just what i have been saying for the last hour |
| 21:56.47 | ceedo | <ceedo> B<^Bceedo^B>^B I think it's clear that the marketplace doesn't select technology based on anything other than price/performance which is my only argument. |
| 21:56.56 | ceedo | was I somehow unclear 59 minutes ago then? |
| 21:57.11 | ceedo | also |
| 21:57.11 | ceedo | <jkk> i am not arguing, i was just saying ceedo's focus on price/performance is misguided |
| 21:57.17 | jkk | ceedo: yes, and i said that is not true |
| 21:57.25 | ceedo | but yet now you admit the market is the reason http is succesful |
| 21:57.36 | {aaron} | there were no other options so price/performance is moot |
| 21:57.44 | {aaron} | s/options/options *in the market*/ |
| 21:57.47 | ceedo | if it cost a million dollars we wouldn't have used it |
| 21:57.47 | jkk | ceedo: well no, i think you don't understand |
| 21:58.12 | jkk | ceedo: the question of price/performance is moot when you are trying to solve a very specific problem |
| 21:58.23 | ceedo | jkk I'll make my statment again: adoption of technology is driven by the marketplace which always chooses based on price/performance from it's available options |
| 21:58.43 | jkk | the question of how, when you have a potentially infinite number of computers out there how do you exchange information between them in a meaningful manner. |
| 21:59.06 | {aaron} | ceedo: i think that is a bit optimistic...we can look at the industry today and see people blow millions on bad software... |
| 21:59.07 | jkk | ceedo: no, what you're saying is a tautology. you cannot say 'the market chooses the best technology' |
| 21:59.14 | jkk | ceedo: that is redundant and self-evident |
| 21:59.23 | ceedo | jkk I didn't say it chose the best software. |
| 21:59.26 | ceedo | http is not the best protocol. |
| 21:59.49 | ceedo | I bet you yourself could design a better protocol |
| 21:59.54 | ceedo | but no one is going to use it. :) |
| 22:00.01 | ceedo | that is all that I am saying |
| 22:00.09 | ceedo | so please don't read other arguments into my core argument. |
| 22:00.26 | jkk | ceedo: you must ask the question 'the best technology for _what_'? and that is why i am saying, you can make a strong argument that http is not the 'best' protocol.. but it is the protocol that makes the browser possible which is crucial |
| 22:00.51 | ceedo | The best technology for whatever the market wants to do. |
| 22:00.56 | ceedo | because the market chooses. |
| 22:01.01 | ceedo | based on price/performance :) |
| 22:01.08 | jkk | ceedo: yes, what you are saying is semantically meaningless |
| 22:01.11 | ceedo | ok. |
| 22:01.25 | jkk | and it does not explain why the market would choose HTTP over say, CORBA |
| 22:01.37 | ceedo | http is cheaper and has inertia. |
| 22:01.43 | ceedo | so it does, I think. |
| 22:01.54 | ceedo | We can also touch on aaron's ubiquitous client standpoint. |
| 22:02.01 | jkk | ceedo: CORBA and its predecesssors had inertia before HTTP |
| 22:02.21 | jkk | ceedo: yes, that is the point.. RPC protocols do now allow for the construction of the 'universal client' |
| 22:02.34 | ceedo | ok |
| 22:02.37 | ceedo | this is gonna go on and on and on |
| 22:02.47 | jkk | yes :) |
| 22:02.51 | jkk | let's stop |
| 22:03.12 | Angus | worse is better. |
| 22:03.12 | ceedo | are you familiar with goodwin's law? |
| 22:03.14 | ceedo | you are a nazi |
| 22:03.19 | jkk | heh |
| 22:03.22 | ceedo | and your protocols remind me of the fascists |
| 22:03.26 | {aaron} | but it cannot be invoked directly to end a discussion :p |
| 22:03.37 | ceedo | yeah unfortunetly that's a true technicality |
| 22:03.50 | ceedo | I was hoping no one would bring that up |
| 22:04.05 | jkk | ceedo: godwin's law can only explain and predict, not produce :) |
| 22:04.12 | Angus | http://www.dreamsongs.com/WorseIsBetter.html |
| 22:04.50 | ceedo | Someone should explain worse is better to the java api people |
| 22:04.58 | Angus | :) |
| 22:05.48 | {aaron} | worse is worse |
| 22:05.50 | jkk | Angus: worse is better is a bad phrase |
| 22:05.51 | ceedo | I think what is interesting and probably outside both of our fields is the huge rift between academic technical development and the business place. |
| 22:05.52 | {aaron} | but faster |
| 22:06.16 | ceedo | business pays for acadamia by proxy in hopes that it will produce a marketable nugget |
| 22:06.24 | Angus | jkk: don't blame me, I didn't coin it. |
| 22:06.31 | ceedo | but doesn't listen to academia when something might be better but not produce more money |
| 22:06.33 | jkk | ceedo: i don't think that's the way it works |
| 22:06.44 | ceedo | no? |
| 22:06.49 | jkk | i think the market is an accident. people happen to like to make money |
| 22:06.53 | Angus | jkk: have you ever read Gabriels's essa on the topic? |
| 22:06.56 | ceedo | Why am I surprised. |
| 22:07.48 | jkk | ceedo: but this does not drive social development. it is like the copyright issue today. even if copyright and patents are completely destroyed, people will still want to make money and will find a way to do it |
| 22:08.20 | {aaron} | http://www.jwz.org/doc/worse-is-better.html |
| 22:08.29 | jkk | ceedo: this is what i tell people when they ask me 'but how will it make me rich?', i say 'i don't care. if you want to be rich, go rob a bank.' |
| 22:08.31 | Angus | jkk: right. They won't publish. The will work entirely in fields where it is possible to keep trade secrets secret. |
| 22:08.52 | ceedo | and so I called that a rift. |
| 22:08.58 | ceedo | And you disagreed :) |
| 22:09.12 | Angus | jkk: the point of patent protection is to further the dissemination of information. |
| 22:09.25 | jkk | Angus: well that is debatable |
| 22:09.39 | {aaron} | that is the point, but it has obviously been perverted |
| 22:09.41 | jkk | nobody really knows what a world without patents would look like |
| 22:09.48 | {aaron} | right |
| 22:09.51 | ceedo | kinda like mexico |
| 22:09.58 | jkk | before patents, the king would grant monopolies ie patents |
| 22:10.33 | Angus | {aaron}: well sure, ou're not supposed to be able to patent things based on prior art, or which are obvious to any skilled practitioner in the discipline. |
| 22:10.52 | {aaron} | Angus: apparently not many of those work at the USPTO |
| 22:11.12 | jkk | Angus: but my point is, you cannot claim that patents make the market possible. the market will always exist because people happen to like to make money |
| 22:11.15 | Angus | {aaron}: I don't think ou should be able to extend patent protection to business plans, for instance. |
| 22:11.25 | {aaron} | there are also things that is it unclear whether it is morally/ethically right to patent: names, DNS, genetic attributes, etc. |
| 22:11.27 | Angus | jkk: I didn't claim that. |
| 22:11.39 | {aaron} | DNS ugh, i meant DNA |
| 22:11.54 | jkk | Angus: because what you are talking about is patents as a form of manipulating the market which is actually, like you pointted out, not what patents were made for (but what they are actually used for) |
| 22:12.02 | {aaron} | a little reform of the USPTO would go a long way |
| 22:12.21 | Angus | jkk: Patents are an attempt to address what has been perceived as a market failure. |
| 22:12.43 | jkk | Angus: no they are not :) |
| 22:12.45 | {aaron} | the whole market itself is a "manipulation"...patents are considered a necessary "manipulation" to advance the progress of arts and sciences - in recent history this has obviously become abused |
| 22:12.59 | jkk | patents are supposed to be purely an incentive mechanism |
| 22:13.10 | Angus | jkk: Namel, that if I develop some new process, and there is no patent protection, it is in my interests not to share an information related to that development, absent patent protections. |
| 22:13.24 | jkk | Angus: nononono :) |
| 22:13.35 | {aaron} | Angus: no...it is not in your interest to *develop it in the first place* |
| 22:13.47 | {aaron} | which is why expirations of patents is so important |
| 22:14.00 | {aaron} | the public is never redeeming the value of these temporary monopolies |
| 22:14.15 | Angus | {aaron}: If I can keep the development a secret, and if it is hard to duplicate without some piece of knowledge that I keep to myself, it is still in my interest to develop it. |
| 22:14.19 | jkk | {aaron}: no you are wrong. people invent stuff all the time which they keep secret and don't patent. patents weren't meant to encourage people to produce new stuff. |
| 22:14.27 | jkk | because people will do that anyway |
| 22:14.46 | ceedo | most big companies give bonuses to employees who patent things. |
| 22:14.48 | {aaron} | jkk: i think they certainly will not if it is too expensive/risky |
| 22:14.58 | ceedo | at Sprint it was worth like a grand |
| 22:15.10 | ceedo | it inspired us to be clever to try to create patentable things |
| 22:15.14 | jkk | in truth, patents are supposed to be a tax on innovation. they are a way for the government to cash in on the creativity of its citizens. |
| 22:15.31 | {aaron} | no...they are not "supposed to be" but have perhaps "become in effect" |
| 22:15.40 | {aaron} | patents are an incentive |
| 22:15.55 | Angus | I agree that the patent office has been allowing people to patent sill stuff. |
| 22:16.26 | Angus | This would be less of a problem if there were stronger penalties against frivolous tortious actions. |
| 22:16.26 | jkk | {aaron}: yeah exactly. but then the real question is, 'do we still need that incentive'? |
| 22:16.59 | ceedo | You know the patent on goretex is up this year. Now everone and their brother are selling gas permiable water barrier fabrics. |
| 22:17.09 | {aaron} | jkk: answer: philosophically debatable, but in reality *yes*, as long as they are *limited* which is what the constitution demands |
| 22:17.20 | {aaron} | ceedo: and that is exactly the way it should work |
| 22:17.34 | {aaron} | ceedo: goretex got their chance, made some money, and now the public benefits |
| 22:17.39 | jkk | {aaron}: you can in fact make a strong argument that the government has no right to grant patents |
| 22:17.41 | Angus | jkk: Why should anyone tie up capital researching something, if the cannot cash in on the research if it leads to something successful? |
| 22:17.54 | ceedo | I just thought I would fan the flames. |
| 22:18.00 | {aaron} | jkk: you could make a strong *philosophical* argument, but in fact the US does have the right, granted in the constitution |
| 22:18.17 | ceedo | gore and eventually dupont of course made a ridiculous fortune on goretex while they enjoyed the patent rights. |
| 22:18.21 | jkk | ceedo: but imagine if the patent on goretex had never been... people would've started selling it much earlier, and there would've been more intense competition leading to the next innovation quicker |
| 22:18.34 | {aaron} | jkk: no, it might not have ever been developed |
| 22:18.45 | jkk | {aaron}: yes and that is the question |
| 22:18.49 | ceedo | if that were true, opensource would produce good software. |
| 22:18.54 | jkk | {aaron}: and of course, that's not true |
| 22:19.11 | Angus | ceedo: open source is heavily subsidized anyway. |
| 22:19.11 | jkk | {aaron}: because something like goretex is so useful it is really inevitable |
| 22:19.23 | {aaron} | hah, i'll refrain from being absolutely certain since there is not way to know in reality |
| 22:19.57 | {aaron} | you could of course argue for *government-sponsored* creation, for things that private individuals or companies refuse to make without patent protection |
| 22:20.04 | Angus | jkk: but how much does it cost to develop it as a product? Why should anone invest that mone if they are forced to give away the end product of that investment? |
| 22:20.12 | {aaron} | i'm not so sure that "goretex" would be high up in the list of items in that case :) |
| 22:20.16 | jkk | {aaron}: in the pre-Information Age, that is in societies where manufacturing is value than patents can be said to make a lot of sense |
| 22:20.33 | {aaron} | jkk: yes, patents on physical things make much more sense |
| 22:20.39 | {aaron} | jkk: never said I agreed with software patents |
| 22:20.54 | jkk | Angus: forced to give away? why would they have to give away the initially secret formula for patent? |
| 22:20.56 | Angus | jkk: um, as far as I can tell manufacturing is still sort of important. I don't actuall live on pure thought. |
| 22:21.09 | jkk | err, formula for goretex |
| 22:21.10 | ceedo | angus hahaha |
| 22:21.29 | {aaron} | patent lifetime and scope needs to be tied to the importance of the accomplishment, the difficulty entailed, and the pace of the technology. period. |
| 22:21.42 | {aaron} | Angus: you mean you are not a bot? |
| 22:22.11 | Angus | jkk: OK, _or_ people would only work on ideas that were hard to imitate, and would not make their research available at all. |
| 22:22.15 | jkk | Angus: which is the real point. at some level, gore must've realized that his accomplishment would inevitably be duplicated |
| 22:22.51 | {aaron} | well, if you higher lots of lawyers it's not inevitable :/ |
| 22:22.52 | jkk | Angus: yes but in reality, there is only one human creation that is truly unique, impossible to imitate, and self-valuing and that is art |
| 22:22.55 | Angus | jkk: Sure, it's a lot easier for someone to imitate something than to develop it in the first place. |
| 22:23.03 | *** join/#tomcat Gnug225 (~Gnug225@24.236.252.196.gha.mi.chartermi.net) |
| 22:23.16 | {aaron} | ugh s/higher/hire/ |
| 22:23.17 | jkk | Angus: everything else, it's just a matter of time before it gets invented and/or imitated |
| 22:23.27 | {aaron} | jkk: i do not agree with that |
| 22:23.40 | {aaron} | jkk: there are tons of things we *can* do, that we simply haven't or won't do |
| 22:24.06 | {aaron} | i dont' believe things will just up and invent themselves in cases where it involves lots of investment and risk |
| 22:24.11 | jkk | Angus: no i don't think you understand. the invention of goretex was not a miracle. if gore had never invented it, it would still have come along. in no way can you say patents encouraged the inital invention because there were many people working on the problem and it was only a matter of time |
| 22:24.18 | Angus | jkk: without patent protections, inventions that were too easy to imitate, but capital-intensive to develop in the first place, would never be funded. |
| 22:24.29 | ceedo | The invention of goretex, was in fact, an accident. |
| 22:24.33 | ceedo | more or less. |
| 22:24.34 | {aaron} | jkk: well, since we don't *know* whether that would have happened, we can only speculate |
| 22:24.44 | ceedo | It was originally cable insulation. |
| 22:24.50 | jkk | Angus: how can you say something is 'easy to imitate but hard to develop'? |
| 22:24.52 | {aaron} | heh |
| 22:25.56 | Angus | jkk: Many things are. It ma take a team of mechanical engineers several years to develop some new mechanism, but guven that development a skilled machinist could duplicate it easil. |
| 22:26.21 | jkk | Angus: do you see where you are heading? patents are not really incentives, they are taxes on innovation. before patents, kings would grant monopolies to some noble.. it was in effect a profit sharing agreement. 'i'll abuse my power to give you a monopoly in return for a cut of the action' |
| 22:26.48 | jkk | Angus: ok so if something is 'hard to invent but easy to imitate' isn't that just like saying it's invention was inevitable? |
| 22:27.13 | jkk | Angus: and if it's so obvious/easy to duplicate... is it really 'novel'? |
| 22:27.13 | {aaron} | i disagree...i think some form of *limited* protection is required to promote the creation of things that would not otherwise be created because they involve too much risk or investment |
| 22:27.23 | Angus | jkk: No. Someone still has to invest the capital for the initial development. |
| 22:27.30 | jkk | {aaron}: and you may in fact be right |
| 22:27.53 | jkk | Angus: but that is not how patents work. they do not reward the first mover. im sure many people were researching goretex before gore was |
| 22:27.56 | {aaron} | jkk: since we there is not simulation for the real world, we might never know |
| 22:28.31 | {aaron} | jkk: they *should* reward the first mover, and that is yet another thing that needs to be reformed |
| 22:28.38 | Angus | jkk: The don't have to reward the first mover to encourage movement. They just have to make it possible that there will be a return on movement. |
| 22:28.47 | {aaron} | but big companies like buying out the small guys on the even of a patent claim anyway |
| 22:29.02 | jkk | Angus: yeah, in political science, that is called a government subsidy |
| 22:29.19 | jkk | but ask yourself who benefits more from a patent, the government or the inventor |
| 22:29.38 | Angus | jkk: Well, it certainl is a _distortion_ of "natural" market mechanisms. |
| 22:30.08 | {aaron} | ha, there is no "natural market" |
| 22:30.13 | jkk | Angus: it is more than a distortion, it is an inversion. |
| 22:30.16 | Angus | jkk: The question is whether or not that distortion creates more good, in addressing market failure, than it does harm in distorting the market. |
| 22:30.42 | {aaron} | in the case of software and other non-tangible items, I believe it does more *harm* |
| 22:31.10 | {aaron} | it's not an all or nothing issue, there is nothing that says we can have patents for some things and not for others, or variable patent duration depending on industry etc. |
| 22:31.21 | Angus | {aaron}: well, I'm not sure that I agree that algorithms should be patentable. |
| 22:31.27 | jkk | Angus: yes... like i said you can make a strong argument for patents but you cannot really say patents drive innovation/market or anything like that. in reality, patents are just another way the government manipulates the market for its own benefit |
| 22:31.40 | {aaron} | Angus: in general i don't think they should be patented |
| 22:32.02 | Angus | jkk: I didn't sa that they did. I said that the argument for them was that they addressed a perceived market failure. |
| 22:32.37 | ceedo | jkk but no one forces people to get patents, and voluntary manipulation is a contradiction. |
| 22:32.45 | {aaron} | jkk: we can't say that because we have no new universe to test in... you are asking for us to "try out" anarchy... the majority of people are not inclined (for whatever reason) to do this |
| 22:32.51 | jkk | Angus: and if, like me, you believe that the only thing that really can be patented is art then it's clear that any patent/limited monopoly system will inevitably be abused since it's based on wrong principles |
| 22:33.22 | jkk | {aaron}: yes but patents make 2 parties richer: the volunteer asking for the patent, and the government granting the patent. everybody else gets screwed |
| 22:33.37 | {aaron} | jkk: wrong...if the patent *expires* then everybody benefits |
| 22:33.39 | jkk | it is like saying corporations 'volunteer' to pollute the environment |
| 22:34.04 | {aaron} | in truth the gov gets (and should get) only a nominal fee for registration of a patent |
| 22:34.16 | {aaron} | there is no reason for the gov. to make money from patent registration |
| 22:34.20 | Angus | jkk: No, not necessaril. If an invention which increases the size of the econom as a whole would not otherwise have been developed, then others may make mone as well, indirectl. |
| 22:34.28 | {aaron} | right |
| 22:34.35 | jkk | {aaron}: no that's not true. there are very few patents that are truly novel. in fact, many patents are filed in such a way to 'maximize impact'.. that is you wait until your competitors are very close to getting where you are, _then_ you file the patent |
| 22:35.04 | {aaron} | jkk: just because that is how it is abused NOW does not mean we should scrap it entirely |
| 22:35.08 | Angus | jkk: But that is an argument about the specifics of our current patent system, as opposed to the general idea of patents. |
| 22:35.23 | {aaron} | yeah |
| 22:35.56 | {aaron} | e.g. in Europe you have to first *prove* no prior art before getting a patent, as opposed to getting one without proving it, and then waiting for challenges in court |
| 22:36.01 | jkk | Angus: yes of course, but it suggests why patents are bad.. that is, if patents don't really drive innovation because such innovation is inevitable, what is the point? |
| 22:36.20 | {aaron} | "if patents don't really drive innovation because such innovation is inevitable," I never agreed to this |
| 22:36.41 | {aaron} | i don't think all forms of innovation are inevitable |
| 22:37.02 | Angus | jkk: But I don't think that that makes the case that those innovations are inevitable. The competitor could be doing that research because the believe that they will be granted the patent. |
| 22:37.32 | Angus | jkk: Another issue is the fundamental research that precedes the development of a product. |
| 22:37.32 | {aaron} | i think there is way way too little evidence to make a sweeping claim that all innovation is inevitable |
| 22:37.46 | {aaron} | yes, tons of money is sunk in research that often goes nowhere |
| 22:37.50 | jkk | Angus: it is like your 'hard to invent but easy to imitate' <--- what does this mean? it makes little sense unless you really just mean 'minimize market exposure' |
| 22:38.15 | Angus | jkk: I don;t entirel disagree with ou, btw. It's just that I think there are likely to be downsides to both positions, and that it makes sense to try to weigh them both. |
| 22:38.46 | {aaron} | we really have no idea what the world would be like without patents so we can't make wild assumptions |
| 22:39.11 | jkk | Angus: i don't entirely disagree with you either. i think something like patents are needed in the same way that the government must subsidize education (public education is another case of market 'distortion)... but i just contest the claim that patents really drive innovation |
| 22:41.15 | jkk | because really the only alternative to patents is to have the government conduct research which we all know is a bad idea :) |
| 22:42.02 | Angus | jkk: No, I think that has real meaning. Think of some novel chemical compund, with all sorts of interesting properties. It might take many chemists many years to do the necessar research to develop it. But in a laboratory it would be eas to analze it, and deduce the necessary production technique. And if it weren't, knowledge of the production technique would never be disseminated. In either case, there is a disincentive toward either developing |
| 22:42.02 | Angus | <PROTECTED> |
| 22:42.48 | {aaron} | and with patents, theoretically it will be researched, and then given to the public (and competitors) after a reasonable amount of time |
| 22:43.10 | {aaron} | unfortunately neither USPTO nor Congress is "reasonable" |
| 22:43.51 | Angus | {aaron}: well, that is a good argument for why we should kep government interference in markets to a minimum. Regulatory capture is inevitable. |
| 22:44.02 | jkk | Angus: if that is the case, then the market doesn't need the compound anyways. that's the point of capitalism, only what's profitable gets executed. if you don't believe that, you have far larger problems than patents |
| 22:44.51 | *** join/#tomcat mod_cure (cure@cs24243192-164.austin.rr.com) |
| 22:46.27 | {aaron} | heh |
| 22:46.39 | {aaron} | go live in a pure capitalist nation then |
| 22:46.54 | {aaron} | with no vaccines, no scientific research, nothing that is immediately profitable...enjoy :) |
| 22:47.02 | Angus | jkk: Well, assuming perfect markets... don't get me wrong, I'm far more likely to argue for market based solutions against them. But there are cases of market failure. As to the degree to which ou can rectify them through government intervention, well that's a can of worms. |
| 22:47.51 | Angus | jkk: and aaron makes another point- our current regulatpor regime imposes artificial costs on the development of certain products. |
| 22:47.53 | {aaron} | the realm of things that are useful to humans is a superset of things that are immediately profitable |
| 22:48.23 | Angus | jkk: pharma products are a good example. Not onl is research expensive, FDA approval is _insanely_ expensive. |
| 22:48.38 | {aaron} | Angus: i believe a pure capitalist market will not provide all things humans typically want in the first place |
| 22:48.43 | jkk | {aaron}: well you don't need to go to such extremes.. in fact i think the real solution is staring us right in the face, ie tomcat ie open source :) but of course, it's a tricky area. the fact of the matter is though, any time you let the government touch the market beyond enforcing contracts it fucks it up hence our current patent situation today |
| 22:48.44 | {aaron} | yes, that is one example |
| 22:49.08 | {aaron} | i don't think the current patent situation today is entirely the government's fault |
| 22:49.21 | {aaron} | it is certainly largely the fault of mega corporations and dirty money |
| 22:49.24 | {aaron} | lobbying, etc. |
| 22:49.26 | Angus | jkk: and pharmaceuticals are an excellent example of something difficult to develop, and steer throough the necessary regulatory proicess, but easy to imitate. |
| 22:49.36 | jkk | Angus: yes, drug research is a special case.. but actually, when you do the numbers, you find that drug research is heavily, heavily subsidized even before patents come into the picture |
| 22:49.55 | Angus | jkk: sure. but that's immaterial to my argument. |
| 22:50.13 | jkk | Angus: and when i say heavily subsdized i mean some corporations basically take drugs out of the research lab of a public university and slap their name on it |
| 22:50.21 | {aaron} | jkk: so you are saying gov subsidization as opposed to patents? that might work, although I think that the gov. is less likely to subsidize "trivial" things that people might want...like "goretex" :) |
| 22:50.22 | Angus | {aaron}: regulatory capture is inevitable. The harder you try to regulate corporations, the dirtier they will become. |
| 22:51.57 | jkk | Angus: but yes, like i said.. something like patents is needed in the same way we need public education.. but i think it is wrong to say patents encourage innovation and i think its questionable how much patents can really help in an information economy |
| 22:53.38 | Angus | jkk: Actually, I'm more bullish on patents than on public education. But I do make a distinction between government funded education, and government run education. |
| 22:54.25 | {aaron} | i don't think it's wrong to say patents encourage innovation...i think they do...the question is "do they encourage innovation that is desirable and would not otherwise occur with a government subsidy system"..I sort of lean to "yes" |
| 22:54.36 | Angus | jkk: I would oprefer to see the government not actually _run_ anything that didn't require the exercise of sovereign power. |
| 22:55.20 | {aaron} | insofar as government defends property rights, government intervention has to exist in the market in some aspect |
| 22:55.38 | Angus | {aaron}: and law enforcement is an exercise of sovereign power. |
| 22:55.43 | {aaron} | physical that is |
| 22:55.45 | jkk | {aaron}: well, there's a difference between enforcing the law and patents |
| 22:55.56 | {aaron} | acknowledged |
| 22:56.01 | Angus | As is the granting of a patent. |
| 22:56.01 | *** join/#tomcat jupi` (jupiter@roswell.warpzone.org) |
| 22:56.15 | jupi` | hello |
| 22:56.30 | {aaron} | hi |
| 22:56.44 | jkk | but of course, the real question is not on the limits of the government, it is on the limits on the market, that is, how much do we trust the market? there are a few people who believe governments are totally uncessary because the market will take care of all human needs |
| 22:57.14 | jkk | and there are people who believe governments are just a compensating factor for the 'few' places where the market fails |
| 22:57.27 | Angus | jkk: I tend to believe that markets are very difficult to establish without at least some government. |
| 22:57.41 | jkk | because arguably, even something like law enforcement could be provided by the market |
| 22:58.56 | {aaron} | i think the former is a terribly naive position |
| 22:59.23 | {aaron} | then again we have never has an experimental anarchy so who knows |
| 22:59.28 | jkk | Angus: i don't really have an opinion. i think the nation-state is inevitable due to the population growth of the species. i also think markets are not 'pre-ordained', they are accidents. so how much one affects the other is really a question of what type of world you want to live in |
| 22:59.33 | Angus | jkk: You could hire private companies to do most of the day-to-day work, but if their authorit to do so, and the rules as to how manner they did so weren't derived from the state, it is difficult to see how that could be considerd law in the sense people normally mean it. |
| 22:59.47 | {aaron} | jkk: i agree |
| 23:00.17 | Angus | jkk: There is a distinction between pre-ordained and effective. I'm not making a natural law argument here. |
| 23:00.20 | {aaron} | Angus: but the market provides choice! if the police murder your family, you are free to "choose" a different police service! :) |
| 23:00.30 | hellabot | "jupi`" pasted JSP Classpath at http://www.hellaweb.com:8888/46 |
| 23:00.54 | jkk | {aaron}: exactly. people will vote with their wallets. |
| 23:00.57 | jupi` | here is my question :o))))))) |
| 23:01.12 | {aaron} | jkk: my point is there are some things that you cannot leave to a "vote" after the fact :) |
| 23:01.22 | ceedo | jupi ssh don't interrupt their argument, I worked hard to get them going. |
| 23:01.42 | jupi` | ? |
| 23:02.00 | {aaron} | embedded tomcat? |
| 23:02.11 | jupi` | The Embedded class |
| 23:02.23 | Angus | Anway, I need to get to work. |
| 23:02.24 | jkk | {aaron}: well that is questionable. you will in fact have a very hard time 'proving' you have a right to life if your death is highly profitable to the market. |
| 23:02.37 | jkk | {aaron}: but yes, i agree :) |
| 23:02.39 | jkk | Angus: later |
| 23:02.49 | jupi` | {aaron}: the Embedded class provided by Tomcat 4.x |
| 23:03.31 | {aaron} | jupi`: ahh...does that take a configuration file, or you programmatically configure it? |
| 23:04.03 | jupi` | it doesn't take a configuration file, but with a digester we can |
| 23:04.25 | jkk | jupi`: you are using Jakarta Digester to configure Embedded? |
| 23:04.31 | {aaron} | maybe i should just use embedded tomcat instead of trying to write my own container |
| 23:04.37 | jupi` | yes |
| 23:04.40 | jkk | {aaron}: yes you should! |
| 23:04.49 | jupi` | {aaron}: but the problem is if we have a classloader hierarchy with a deep > 3 |
| 23:04.59 | jkk | writing your own container == bad |
| 23:05.31 | {aaron} | jkk: is there a significant performance/memory savings? i was hoping to get something running in java on a very old machine... |
| 23:05.32 | jupi` | {aaron}: because the WebappLoader.setClasspath method take for the JSP classpath only the third parent classloader |
| 23:05.51 | {aaron} | jkk: i don't mind if it is "bad" as long as it works...it is just for my own play :) |
| 23:06.00 | jupi` | {aaron}: so the jspclasspath is missing some URLs and JSP can't compile |
| 23:06.13 | jkk | jupi`: hmmmm...say that again? you basically need all of tomcat's libs |
| 23:06.32 | jupi` | yes |
| 23:06.36 | jkk | read http://www.onjava.com/pub/a/onjava/2002/04/03/tomcat.html ---> for how to embed tomcat |
| 23:06.41 | {aaron} | thanks |
| 23:07.59 | jupi` | there is no problem for me for using Embedded, this is when it use JSP and compile them, if my lib is in the first classloader and there is a deep of 5 in the classloader hierarchy, the tomcat method doesn't add my url/.jar to the jsp classpath and it won't compile |
| 23:08.46 | *** join/#tomcat TomekN (~tnowak@zorro.netventure.pl) |
| 23:08.50 | TomekN | hello. |
| 23:08.56 | jupi` | because it lookup only on a deeper = 3 |
| 23:09.01 | jupi` | and why 3 and not 4 or more |
| 23:09.07 | {aaron} | jupi`: add fake classloaders maybe? |
| 23:09.14 | jkk | jupi`: how do you know it only looks 3 up? |
| 23:09.20 | {aaron} | it sees <= 3 or >= 3 |
| 23:09.21 | {aaron} | ? |
| 23:09.31 | jupi` | because it is in the source code |
| 23:09.36 | jkk | jupi`: where? |
| 23:09.50 | jupi` | WebappLoader, setClasspath method |
| 23:10.14 | jkk | i would hope the jakarta developers aren't that dumb... there is some silly code in tomcat but why looking only 3 up instead of always delegating makes no sense |
| 23:10.18 | {aaron} | while ((layers < 3) && (loader != null)) { |
| 23:10.25 | jupi` | yes |
| 23:10.43 | jupi` | but why 3 and not only while (loader != null) |
| 23:10.47 | jkk | jupi`: you should file that as a bug and complain on tomcat-dev then |
| 23:11.02 | jkk | i can't think of any reason to not always delegate to the parent classloader |
| 23:11.18 | jupi` | because there is no explanation of the "3" number |
| 23:11.32 | jkk | yeah it sounds like a classic hack |
| 23:11.33 | {aaron} | maybe they don't want to look in bootclasspath? |
| 23:11.40 | {aaron} | or classpath |
| 23:11.57 | jupi` | yes but when we embed tomcat |
| 23:12.02 | {aaron} | i know |
| 23:12.05 | {aaron} | i understand |
| 23:12.37 | jkk | it's a shame that the jasper developers would write such container unfriendly code. |
| 23:13.39 | {aaron} | apparently they inspect a context attribute to determine classpath |
| 23:13.45 | jupi` | yes |
| 23:13.56 | {aaron} | it must be so they can launch an external javac if necessary |
| 23:14.07 | jkk | or you could just not use jsp which isn't the greatest technology either :) |
| 23:14.17 | jkk | {aaron}: why would they need an external javac? to compile jsps? |
| 23:14.52 | {aaron} | i don't know why, but I think that may be an option |
| 23:15.00 | jupi` | but in 4.1.x they use facade and so, if we set the Globals.CLASSPATH_ATTR attributes, this is not used as we set only to a copy of the context |
| 23:15.03 | {aaron} | for one thing, it has been a long standing bug that embedded javac leaks memory |
| 23:15.05 | *** join/#tomcat Archeron (archeron@archeron.wavefire.com) |
| 23:17.29 | jupi` | so if I want an answer, I must post a bug or say it to mailing list |
| 23:17.48 | jkk | jupi`: well you could change the source code, recompile and hope for the best |
| 23:17.54 | {aaron} | copy WebappLoader, modify it, and setLoader() |
| 23:17.57 | jkk | but really it is a bug and you should complain |
| 23:18.33 | jupi` | yes but I want to use embedded, I don't want to say people which use my program to recompile their catalina class |
| 23:18.57 | {aaron} | jupi`: create your own WebappLoader |
| 23:19.05 | {aaron} | copy + paste + modify + setLoader() |
| 23:19.13 | jkk | jupi`: you can redistribute tomcat with your app or yes, write your own webapp loader |
| 23:19.27 | jupi` | yes but the WebappLoader is created with the StandardContext class |
| 23:19.43 | jkk | jupi`: then write your own Context implementation |
| 23:19.45 | jupi` | so I must modify or extends this 2 class |
| 23:19.52 | jupi` | only to remove a 3 |
| 23:19.57 | jupi` | :) |
| 23:20.08 | jupi` | This is not a proper integration |
| 23:20.28 | jkk | jupi`: well consider if tomcat was closed source, you wouldn't even know what the problem is |
| 23:20.46 | jupi` | I don't blame tomcat |
| 23:20.57 | jupi` | or open source or closed source :) |
| 23:21.31 | jupi` | I just want to know why a 3, which we can't change or set with server.xml or something like that |
| 23:22.22 | jkk | jupi`: i dont think there is a good reason unless perhaps it's mandated by the jsp spec which i've never read but highly doubt. it smells like a bug |
| 23:23.34 | jupi` | yes I think too |
| 23:24.49 | jkk | jupi`: but that is why they call it a 'workaround' |
| 23:25.16 | jupi` | :o) |
| 23:39.04 | jupi` | good nite |