|
all i have gotten out of my rear end backwards nightmare is a drinking problem
|
# ? Apr 13, 2016 00:15 |
|
|
# ? May 12, 2024 00:59 |
|
Bloody posted:all i have gotten out of my rear end backwards nightmare is a drinking problem I wouldn't really call it a "problem"
|
# ? Apr 13, 2016 00:52 |
|
Bloody posted:all i have gotten out of my rear end backwards nightmare is a drinking problem going to a bar this weekend with one of my qaers where i will be getting a beer flight that is 25 4oz glasses of craft beer on a plank. this is a normal way to spend a saturday
|
# ? Apr 13, 2016 01:02 |
|
well you're in Wisconsin what else is there
|
# ? Apr 13, 2016 01:03 |
|
Bloody posted:well you're in Wisconsin what else is there lots of music in Madison, and sports, but we get really drunk for those too so
|
# ? Apr 13, 2016 01:08 |
LeftistMuslimObama posted:y know, i don't think i did. it's not too hard and ive been at work since 6am so i guess ill take a break and write one now Every MUMPSpost is fascinating. Thanks for the write-up!
|
|
# ? Apr 13, 2016 01:23 |
|
when i read the mumps post the paragraph descriptions sounds almost reasonable (in an old language kind of way) and then i see the code examples and my brain just nopes out
|
# ? Apr 13, 2016 01:35 |
GrumpyDoctor posted:when i read the mumps post the paragraph descriptions sounds almost reasonable (in an old language kind of way) and then i see the code examples and my brain just nopes out Honestly MUMPS doesn't seem *that* much worse than FORTRAN 77, which I occasionally have to work with. That is, it seems like it pre-dates modern conventions (which are there for good reason), but it isn't necessarily horrible by the standards of its time.
|
|
# ? Apr 13, 2016 01:48 |
|
i've done some amount of physics in college, which obviously means i did end up touching fortran 77 less than a decade ago. i don't remember it being anywhere close to what mumps seems to be in terms of being bad, but then it was all for numerical simulation code which is its strength, and just for fairly simple stuff, so maybe i would have a different opinion if i had dealt with giant codebases of the stuff instead
|
# ? Apr 13, 2016 02:08 |
|
uncurable mlady posted:I think TDD is a good idea wrapped up in a bunch of dumb dogma. the kernel of "write tests against spec, then write function until tests pass" is a pretty useful maxim and something anyone can glom onto even if they don't really have experience with the unit test framework. tdd is great precisely because it forces a level of respect for tests, though youre right that trying to convince a dev to start writing tests is an unreasonably challenging task. im curious though, how do the devs gently caress with your test framework? your typical xUnit framework is pretty straightforward and im having a hard time imagining how people gently caress it up other than retaining state so your execution order matters.
|
# ? Apr 13, 2016 02:11 |
|
Bloody posted:well you're in Wisconsin what else is there Cheese. I'm in an OS class where the majority took all the Java classes, so for an insane reason the prof lets people choose whether to write programs in C or Java. Anyway, today a classmate came up to me and asked about how to do threading, and when I replied that I don't use Java he asked how I made a thread to "capture a class". When I told him I just used the POSIX lib and didn't need to set up a class, just call the function to work on the data instead, he looked at me like a deer in headlights. I don't know who is the most terrible programmer here.
|
# ? Apr 13, 2016 02:32 |
|
mekkanare posted:Cheese. lol, c was mandatory in my OS class and we implemented threading from scratch on a kernel that didnt have threading support. your classmate would have been hosed. also, the actual language mumps is terrible. the strength, if there is one, is that b-trees are a native data structure with extremely efficient builtin functions for traversing them. nowadays, that's still not a a great reason to pick it if you're starting from scratch, but in the 70s that was a huge win for its specific use case, and theres not a compelling reason to force all our customers to migrate their core systems to a new platform. plus we're singlehandedly keeping a lot of greybeard cache/unix admins employed
|
# ? Apr 13, 2016 02:48 |
|
YeOldeButchere posted:i've done some amount of physics in college, which obviously means i did end up touching fortran 77 less than a decade ago. i don't remember it being anywhere close to what mumps seems to be in terms of being bad, but then it was all for numerical simulation code which is its strength, and just for fairly simple stuff, so maybe i would have a different opinion if i had dealt with giant codebases of the stuff instead the way i remember it is if youre not trying to do anything fancy with things like "input and output" and "talking to things" fortran77 is ok for numerics
|
# ? Apr 13, 2016 02:53 |
|
uncurable mlady posted:I think TDD is a good idea wrapped up in a bunch of dumb dogma. the kernel of "write tests against spec, then write function until tests pass" is a pretty useful maxim and something anyone can glom onto even if they don't really have experience with the unit test framework. it's pretty simple if you really think about it. write test stubs based on spec. if you want to do tdd, write your tests first and then try to make them green. if you don't, write your code first then write tests that follow the description in the stubs. qa is designed to test workflows and behaviors, not code. in summary, write tests. write all the tests.
|
# ? Apr 13, 2016 03:03 |
|
shaggar, you do emr stuff. write tests.
|
# ? Apr 13, 2016 03:03 |
|
Seriously, we built a unit test framework for loving mumps. You can write tests for whatever poo poo you're using.
|
# ? Apr 13, 2016 03:09 |
|
LeftistMuslimObama posted:lol, c was mandatory in my OS class and we implemented threading from scratch on a kernel that didnt have threading support. your classmate would have been hosed. goddamn, where were those classes in my college? i would have loved to do that. all my cs classes were worthless down to some of my profs straight teaching the wrong definition of things i still don't feel like i know the right words for stuff
|
# ? Apr 13, 2016 03:12 |
|
Blinkz0rz posted:in summary, write tests. write all the tests.
|
# ? Apr 13, 2016 04:04 |
|
JimboMaloi posted:tdd is great precisely because it forces a level of respect for tests, though youre right that trying to convince a dev to start writing tests is an unreasonably challenging task. im curious though, how do the devs gently caress with your test framework? your typical xUnit framework is pretty straightforward and im having a hard time imagining how people gently caress it up other than retaining state so your execution order matters. our test framework uses xUnit but we have a broad 'functional' testing framework that essentially is a XML DSL that can run arbitrary code or scripts or selenium that was built by interns years ago and it's exactly as bad as you might imagine send booze
|
# ? Apr 13, 2016 04:19 |
|
also I just spent like 3 hours refactoring 4 lines of code in aforementioned framework and I'm only pretty sure I didn't break anything so I definitely belong itt
|
# ? Apr 13, 2016 04:20 |
|
uncurable mlady posted:also I just spent like 3 hours refactoring 4 lines of code in aforementioned framework and I'm only pretty sure I didn't break anything so I definitely belong itt honestly, ive always found that the hardest/most time consuming work involves tiny chunks of code, because typically if youve zeroed in on that tiny chunk its doing something loving stupid that cant be fixed in a straightforward way because half your codebase is broken by the stupid and the other half expects it and implements workarounds that are broken when you unfuck the code.
|
# ? Apr 13, 2016 04:27 |
|
uncurable mlady posted:send booze yosmas starting early this year
|
# ? Apr 13, 2016 05:01 |
|
uncurable mlady posted:our test framework uses xUnit but we have a broad 'functional' testing framework that essentially is a XML DSL that can run arbitrary code or scripts or selenium that was built by interns years ago and it's exactly as bad as you might imagine do you have the story on how that came to exist?
|
# ? Apr 13, 2016 05:05 |
|
hello XML DSL test framework buddy how did it come about? not very carefully
|
# ? Apr 13, 2016 05:31 |
|
lol java script: http://www.reapp.io/ An easier, faster way to build apps with React and JavaScript. vs https://github.com/reapp/reapp NOTICE: Needs contributors, current broken, don't open a ticket open a PR I swear people are making the splash site for their library well before they worry about getting it stable.
|
# ? Apr 13, 2016 05:56 |
|
HoboMan posted:so what the gently caress is DevOps? it's like agile. it means whatever the gently caress you want it to mean. it means both different from the status quo and also the status quo. so, agile means "we can change the project overnight and the rest of you need to pick up the slack", devops means "well, someone's gotta take the pager, well, i guess it's you, dev" or sometimes devops means "well, we automated installing the ruby app, but we haven't worked out how to turn it on an off automatically" here's some material i'm trying out: mazlow's heirarchy of operations (but it's upside down) - snowflakes (production is special and unique. good luck) - playbooks (copy and paste to install rather than googling) - pets (pre-existing and shared production, staging, manual provisioning) - builds (make me an X) - configuration (change X, change all X, or 'parallel ssh') - reporting (state changes, metrics for throughput, latency, cost, and event/change logs) - recovery (high availability, backups, restore, rewind, replay, merge) - provisioning (api, replicas, multi tenanancy, formations) - platform-as-a-service (notifications, maintenance windows, plans, configuration, topology, policy) - managed platform (the same, but you pay someone else to do it) in some ways devops is trying to push towards 'managed platforms', but most of the time we stop around 'parallel ssh' and hack the rest of the poo poo up tef fucked around with this message at 06:03 on Apr 13, 2016 |
# ? Apr 13, 2016 05:59 |
|
HoboMan posted:this is the first thing i looked at after google "tef blog": http://programmingisterrible.com/ it's me Bloody posted:all i have gotten out of my rear end backwards nightmare is a drinking problem "programming is terrible, lessons learned from a life wasted" that second part is an in-joke about how self destructive i am ha ha
|
# ? Apr 13, 2016 06:05 |
|
to save you some time, here are the only things that people have linked to http://programmingisterrible.com/post/65781074112/devils-dictionary-of-programming this did the rounds and i hear the 'framework' definition occasionally http://programmingisterrible.com/post/139222674273/write-code-that-is-easy-to-delete-not-easy-to this is basically a very edited yopost
|
# ? Apr 13, 2016 06:07 |
|
Maluco Marinero posted:https://github.com/reapp/reapp Last real commit 7 months ago, no pull requests, but 3,346 stars and 174 forks? what the gently caress is going on in there?
|
# ? Apr 13, 2016 06:12 |
|
Asymmetrikon posted:Last real commit 7 months ago, no pull requests, but 3,346 stars and 174 forks? what the gently caress is going on in there? No idea. Only got told to look at it because we're do some settingsy modal stuff, but I lolled it off pretty quickly. Javascript frameworks never change.
|
# ? Apr 13, 2016 06:30 |
|
it's cool to get paid to work on open source stuff
|
# ? Apr 13, 2016 06:38 |
|
triple sulk posted:it's cool to get paid
|
# ? Apr 13, 2016 07:05 |
|
LeftistMuslimObama posted:lol, c was mandatory in my OS class and we implemented threading from scratch on a kernel that didnt have threading support oh, lucky you, our OS class was all about fork and sysv ipc, the project was about writing a silly toy task dispatcher. in our study group we called it bogotron 2000 because it did a lot of work to accomplish nothing . I had the honor of having to explain the name to the professor lol
|
# ? Apr 13, 2016 09:36 |
|
Bloody posted:lol if u cant just leave when ur sick of the poo poo we use punchcards. well not literally, but a web form where you have to click on a button when you arrive and leave. this is separate from the timesheets where you still need to go type your hours you can be drat sure that i loving beeline for the login button as soon as i come in from the front door. and after i click the logout button, i leave behind a cartoonesque dust cloud where i was standing.
|
# ? Apr 13, 2016 11:40 |
|
Maluco Marinero posted:Javascript frameworks never change. the exact opposite, really
|
# ? Apr 13, 2016 11:57 |
|
JimboMaloi posted:
probably the same way it comes about everywhere else, organically. some years ago, when the company was still a fledgling, people decided that we needed a way for the nascent QA department (which was like 3 people) to write tests. our product is rather complex, and wasn't written with testability in mind, so you had to actually deploy it to really make sure things were working. some interns started working on a project that would allow for the automation of standalone utilities (like an application that populated the app with data, etc.) into a series of tests. it wasn't terribly complex at first, but over time, it grew. now it's a morass of stuff that isn't really well understood even by the two of us that work on it, there's random code everywhere that never executes and there's no real explanation why (c.f., one of the core classes has a run and a runFast method that seem to do the same thing, but only run is ever called. why does runFast exist? we've never bothered trying it out). the real problem that it causes is that since it's all the QA people know how to use, they try to fit everything into a Selenium RC shaped box, even when what they're trying to test could be accomplished via calling APIs through a script or something. every time they do something in Selenium RC, it adds to our 8000+ module backlog of selenium modules that we want to port to a webdriver framework. Unfortunately, the webdriver framework is somewhat complicated in part by it being the product of someone who is now a product dev and is honestly a pretty clever guy. The code for the framework is, unexpected, also very clever. Clever code is a loving PITA to maintain or understand what the gently caress it's doing. the really hilarious part of this is that one of the reasons there's so much cruft and bullshit is that for the longest time (literally years) they had developed all sorts of hacky-rear end workarounds to deal with the fact that a potential configuration of the product in the wild involved using AD federation to an external user store. out of the box, the EUS ADFS would try to do basic auth when you logged in. since basic auth popups aren't managed by the browser, selenium couldn't interact with them. this lead to thousands of LOC written in order to use AutoIT to enter a username/password in that popup and click 'OK'. imagine everyone's surprise when I noted that you could change the behavior of the external user store by editing the web.config and changing one line to tell it to present a signin page rather than a basic auth popup...
|
# ? Apr 13, 2016 13:17 |
|
redleader posted:over the last couple of days i wrote a truly hideous sql function that does some annoying business logic stuff. it's going to be a maintenance nightmare for the poor bastard who has to deal with it later. i mean, i wrote the drat thing and i'm not sure that i can fully explain all of it. it sort of... evolved organically, in that i understood each change to the function individually, but if you asked me to walk you through the whole thing i'd have to hand wave and mumble a lot to absolutely no one's surprise, i had to almost totally rewrite that awful sql function because the client's requirements weren't what we thought they were it's slightly better now, so that's nice
|
# ? Apr 13, 2016 13:23 |
|
Wheany posted:we use punchcards. well not literally, but a web form where you have to click on a button when you arrive and leave. lol if you havent written a program that "clicks" the button at 9 AM and again at 5:30 PM M-F for the rest of eternity
|
# ? Apr 13, 2016 14:35 |
|
Bloody posted:lol if you havent written a program that "clicks" the button at 9 AM and again at 5:30 PM M-F for the rest of eternity look at this amateur who doesn't slightly randomize the times by +/- 5 mins so it doesn't seem out of place
|
# ? Apr 13, 2016 14:36 |
|
|
# ? May 12, 2024 00:59 |
|
today in terrible programmerism, verilog is a horrendous tightly-coupled pile of rear end so i have to "refactor" a gigantic monolithic mess into a slightly different gigantic monolithic mess refactor in this case probably means "delete and start over"
|
# ? Apr 13, 2016 14:36 |