|
Nomnom Cookie posted:that guy is editing one script because he doesn't know any better. it's also A Thing to concatenate and minify the js for a page, it speeds js up lol javascript execution speed is not at all affected by combining/minifying scripts. the speedup is because you're not getting the overhead of taking down and negotiating new http requests. this has nothing to do w/ the language itself.
|
# ? Oct 8, 2013 17:16 |
|
|
# ? May 23, 2024 15:46 |
|
Tiny Bug Child posted:how many people work here? probably about 30-40 full timers. idk it's started to hit the point where i don't actually know everyone who works over in the spain office 1 mil / mo on 40 ftes ain't shabby i'm in the wrong business
|
# ? Oct 8, 2013 17:22 |
|
spongeh posted:javascript execution speed is not at all affected by combining/minifying scripts. the speedup is because you're not getting the overhead of taking down and negotiating new http requests. this has nothing to do w/ the language itself. the js language standard is what mandates the lovely serial downloads and serial loads all other http resources are pulled in parallel
|
# ? Oct 8, 2013 17:22 |
|
Notorious b.s.d. posted:the js language standard is what mandates the lovely serial downloads and serial loads nope
|
# ? Oct 8, 2013 17:25 |
|
please explain to me how you can implement the standards-mandated dom without the lovely serial loads i'm all ears about why mozilla and chrome choose to be slow as gently caress for no reason
|
# ? Oct 8, 2013 17:28 |
|
javascripts are pulled the same way all other resources are, its the execution that's done sequentially. the reason people minify their javascript libraries is because they are laughably huge and its less bandwidth for the server
|
# ? Oct 8, 2013 17:28 |
|
the reason is that http spec says you shouldnt have more than 2 simultaneous persistent connections
|
# ? Oct 8, 2013 17:30 |
|
Notorious b.s.d. posted:public, static, and main are all fuckin fine. get a grip. they're well-specified, they are well-understood by users coming from other languages, and they provide useful features it was a joke, that was what pollyanna as blubbering about in the coding horrors thread after reading babby's first java tutorial for 3 minutes
|
# ? Oct 8, 2013 17:31 |
|
Posting Principle posted:the reason is that http spec says you shouldnt have more than 2 simultaneous persistent connections no browser hews to this hth
|
# ? Oct 8, 2013 17:32 |
|
spongeh posted:javascript execution speed is not at all affected by combining/minifying scripts. the speedup is because you're not getting the overhead of taking down and negotiating new http requests. this has nothing to do w/ the language itself. startup time is affected and that's what I mean. it's part of speed. although i wouldn't be surprised if minified js ran faster too
|
# ? Oct 8, 2013 17:44 |
|
Notorious b.s.d. posted:public, static, and main are all fuckin fine. get a grip. they're well-specified, they are well-understood by users coming from other languages, and they provide useful features idk i mean yes that's dumb and stupid. but it shouldn't be costing anyone days of work (jar hell) or severely limiting the types you can create (no structs, type erasure). also final and const are different, final is const pointer only
|
# ? Oct 8, 2013 17:49 |
|
jar hell isn't a thing if you're using maven and u aren't using maven ur a moron. type erasure is not really a big deal and is only slightly annoying when you want to do something that's probably 90% of the time a bad idea.
|
# ? Oct 8, 2013 17:52 |
|
Shaggar posted:jar hell isn't a thing if you're using maven and u aren't using maven ur a moron. jar hell is a thing in j2ee, because classloader ordering/behavior is poorly specified but if you use j2ee you probably hate your job for lots of other reasons
|
# ? Oct 8, 2013 17:54 |
|
Nomnom Cookie posted:idk i mean yes that's dumb and stupid. but it shouldn't be costing anyone days of work (jar hell) or severely limiting the types you can create (no structs, type erasure). also final and const are different, final is const pointer only yeah having final mean a bunch of things is not going to cost you a lot of time over your career. but it's dumb and stupid and hard to learn, and java was meant to be clear and simple and easy to learn
|
# ? Oct 8, 2013 17:55 |
|
its not a problem unless ur doing dumb things or the server is buggy.
|
# ? Oct 8, 2013 17:55 |
|
always use osgi
|
# ? Oct 8, 2013 17:59 |
|
<script async> bitches
|
# ? Oct 8, 2013 18:09 |
|
Shaggar posted:jar hell isn't a thing if you're using maven and u aren't using maven ur a moron. it's a thing if not all of your dependencies are properly mavenized, and it's not always feasible to mavenize them yourself
|
# ? Oct 8, 2013 18:11 |
|
ooh can i play
[#]for how much of a selling point it is, the concurrency story is really not that impressive, especially integration with io, so far. [#]same as above but about C interop. like, haskell probably does both of these better or at least more conveniently. [#]unrelated but i forgot how to do numbered lists in bbcode [#]compile times are probably not going to go anywhere because of the large compilation units
|
# ? Oct 8, 2013 18:21 |
|
Nomnom Cookie posted:it's a thing if not all of your dependencies are properly mavenized, and it's not always feasible to mavenize them yourself ya it is
|
# ? Oct 8, 2013 18:22 |
|
get better deps
|
# ? Oct 8, 2013 18:24 |
|
if you've got stuff you know is on the server, add it as provided in ur pom (or better yet create a pom w/ all the provideds listed and make that a reusable dependency). if the stuff is custom crap, shove it into nexus
|
# ? Oct 8, 2013 18:27 |
|
pom pom wei wei wei
|
# ? Oct 8, 2013 18:27 |
|
Nomnom Cookie posted:it's a thing if not all of your dependencies are properly mavenized, and it's not always feasible to mavenize them yourself it's really not that hard, if you have the source, just add a pom if you only have a jar/war/ear, just use the nexus to upload the artifact directly, with a bespoke pom make sure your versioning scheme matches theirs, and what i always do is put my org's group name and extend the artifact name to include their group name (or what i would assume is their group name), spring does this so like, if you have jar "lovely thing" from oracle and it's version 1.2.3, and your org is something awful the gav should be com.somethingawful/com.oracle.lovely-thing/1.2.3
|
# ? Oct 8, 2013 18:29 |
|
Vanadium posted:[#]unrelated but i forgot how to do numbered lists in bbcode list=1
I was going to use a pre for the formatting but lol, it renders even in a pre block: pre:
|
# ? Oct 8, 2013 18:30 |
|
trex eaterofcadrs posted:it's really not that hard, which works great until a coworker decides to run his artisanal hadoop jar on my classpath and everything is hosed and i have to either support his attempts to fix it or take over packaging his mapreduce jobs i'm not saying it's a problem with no solutions, i'm saying it's a problem and you have to solve it. it sucks even if you know what you're doing. it sucks to look at a list of 300+ jars to spot conflicts. it sucks integrating dependencies that bundle their deps, have old deps, and don't have an unbundled option p.s. hadoop packaging is so terrible. its just the worst
|
# ? Oct 8, 2013 18:49 |
|
the solution is use maven and fire the people who don't use the solution
|
# ? Oct 8, 2013 19:08 |
|
PleasingFungus posted:list=1 Use code instead of pre code:
|
# ? Oct 8, 2013 19:16 |
|
Nomnom Cookie posted:which works great until a coworker decides to run his artisanal hadoop jar on my classpath and everything is hosed and i have to either support his attempts to fix it or take over packaging his mapreduce jobs yeah so your problem is not that mavenizing an unmavenized jar is hard. it's definitely not hard. it takes about 60 seconds. no the problem is that you have some rear end in a top hat jars that bundled their deps. which means you have to unpack and repack the jar. which is not that hard in itself, it just makes me really, really angry when i have to do it
|
# ? Oct 8, 2013 19:18 |
|
the only thing worse than bundling your deps in the jar is loving it up with OneJar that way you can have bundled deps AND a hijacked classloader that ignores your classpath
|
# ? Oct 8, 2013 19:19 |
|
Nomnom Cookie posted:p.s. hadoop is so terrible. its just the worst
|
# ? Oct 8, 2013 19:32 |
|
Nomnom Cookie posted:which works great until a coworker decides to run his artisanal hadoop jar on my classpath and everything is hosed and i have to either support his attempts to fix it or take over packaging his mapreduce jobs i thought hodapp was just something you ran on the mapred cluster and not on your workstation
|
# ? Oct 8, 2013 19:59 |
|
Cocoa Crispies posted:i thought hodapp was just something you ran on the mapred cluster and not on your workstation ok technically they took my classpath and shoved ALL OF IT into hadoop at oldjob i started thinking the only sane thing was to make everything a WAR and run it inside tomcat. this is what newjob does
|
# ? Oct 8, 2013 20:14 |
|
Notorious b.s.d. posted:yeah so your problem is not that mavenizing an unmavenized jar is hard. it's definitely not hard. it takes about 60 seconds. oh yeah if its a little jar mavenizing it is ez. the problems come from projects like hadoop that just assume they own the classpath and bundle up a whole bunch of poo poo and provide zero guidance on what deps are actually necessary and what you can remove. storm does this too it's not like this is a new problem, servlets addressed it a long rear end time ago by giving each war its own classloader. but open sores is poop from a butt
|
# ? Oct 8, 2013 20:19 |
|
so either run each hadoop in its own jvm or run it inside its own container. ez pz problem solved.
|
# ? Oct 8, 2013 20:38 |
|
Nomnom Cookie posted:oh yeah if its a little jar mavenizing it is ez. the problems come from projects like hadoop that just assume they own the classpath and bundle up a whole bunch of poo poo and provide zero guidance on what deps are actually necessary and what you can remove. storm does this too storm and scalding are both way worse than hadoop a normal java application has properties files, config files, and shell scripts to configure and boot the drat thing. in the worst case, you get like jetty, where the properties/config/shell just launches start.jar, which relaunches the jvm storm and scalding turn this poo poo on their head. storm uses python scripts + yaml config + environment variables to launch. scalding uses ruby. it's impossible to turn a storm or scalding app into a normal deployable thing; no matter what you do you're gonna come back to their wretched scripting yeah we know java has 10 years of deployment tooling but we thought "python storm.py deploy-my-poo poo" was somehow better
|
# ? Oct 8, 2013 20:44 |
|
also at least the storm python scripts are halfway readable scalding is so bad i just couldn't get it to run on my desktop; it like wanted scala 2.8 and ruby 1.8 and prayer. and there were no usable docs just example code, which of course requires scalding to already be working it's like they just dumped somebody's source tree + readme.md to github and called it done
|
# ? Oct 8, 2013 20:45 |
|
theres probably a maven plugin that handles generating the startup scripts and prob windows java service wrapper installation scripts too
|
# ? Oct 8, 2013 20:46 |
|
Examine your problem. Is the problem solved by using maven/maven plugins? if the answer is no, your problem is you.
|
# ? Oct 8, 2013 20:48 |
|
|
# ? May 23, 2024 15:46 |
|
Notorious b.s.d. posted:also at least the storm python scripts are halfway readable sounds real bad e: woah big
|
# ? Oct 8, 2013 20:49 |