|
in c f(void) explicitly says f doesn't take arguments. in c++, saying f() means the same thing as f(void) tho
weird fucked around with this message at 16:11 on Apr 11, 2016 |
# ? Apr 11, 2016 16:09 |
|
|
# ? May 25, 2024 14:13 |
|
i was going to say something like "but then the linker is gonna tell you to gently caress off won't it?" but i realized that 1) the linker will have access to a malloc definition anyhow, and 2) there's about 0% chance that a c linker will care about return types not matching awesome
|
# ? Apr 11, 2016 16:10 |
|
YeOldeButchere posted:i was going to say something like "but then the linker is gonna tell you to gently caress off won't it?" but i realized that 1) the linker will have access to a malloc definition anyhow, and 2) there's about 0% chance that a c linker will care about return types not matching C function name mangling doesn't encode arguments, return type or anything else, just the name, traditionally with an underscore in front to distinguish it from symbols declared in asm. calling conventions for C functions are designed so that if you pass too many arguments, the extra arguments will be ignored, and if you pass too few arguments, the missing arguments will be filled with random garbage, but unless the garbage makes the function crash, the program will run fine. it will probably have meaningless behavior (garbage in garbage out), but the wrong number of arguments will not by itself crash the program or compromise its integrity C is "awesome"
|
# ? Apr 11, 2016 16:22 |
|
yeah, i sorta knew about most of that stuff, partly because of c++ compilers throwing out a bunch of deprecated c behavior errors about implicit declarations when misspelling function names, and partly thanks to the Underhanded C Contest where mismatching declarations/definitions are not an uncommon entry point for hidden behavior
|
# ? Apr 11, 2016 16:36 |
|
ahmeni posted:remember that programming is a discipline and that you should write some loving tests, where are your goddamn tests man this is a good question, i just came out of a meeting to outline the plan for people to start writing regression tests (no, we don't have any right now). it's not me that's gotta write them, we apparently do have a qa team. i mentioned that automating regression testing instead of doing them by hand every time would probably be a good idea. my boss replied "ok, you should go get on implementing that then". gently caress, uhhhhhhhhhhhh??? Bloody posted:or at least acknowledge you should write tests then sigh wistfully and get back to implementing new features without fixing old bugs also this. there's so many tickets. we're supposed to be "feature locked" for the next release, but haha no HoboMan fucked around with this message at 16:56 on Apr 11, 2016 |
# ? Apr 11, 2016 16:50 |
|
HoboMan posted:my boss replied "ok, you should go get on implementing that then". gently caress, uhhhhhhhhhhhh??? no good deed goes unpunished
|
# ? Apr 11, 2016 18:54 |
|
scratch your own itch. document your current workflow to a level that someone could take over just from reading it, then work on improving the obvious bits. automate one bit of the regression test every week. week 1: make kicking it off pushbutton. week 2: make collecting results pushbutton. week 3: triage, week 50: sippin mojitos while HoboBot does all the work
|
# ? Apr 11, 2016 19:03 |
|
JawnV6 posted:scratch your own itch. document your current workflow to a level that someone could take over just from reading it, then work on improving the obvious bits. , but made out of poo poo
|
# ? Apr 11, 2016 19:23 |
|
so anyone recommend a documentation and testing setup for visual studio 2010? e: also deployment because i currently deploy by copy-pasting the files onto the server HoboMan fucked around with this message at 19:45 on Apr 11, 2016 |
# ? Apr 11, 2016 19:40 |
|
go hit up Ithaqua in the grey forums, he does that for a living but my recommendations are: 1. Open up notepad and quit pretending you need a fancier tool
|
# ? Apr 11, 2016 19:58 |
|
btw, I'm on one end of the pendulum swing where I'm avoiding building tools. wondering if anyone else hits this point i had to set up 60+ server entities. i scripted half of it, then did a naive check on every single one of them and manual work to finish them up. the tool for the second half would not have had great ROI, but I also jump to "i know! ill build a TOOL to do it!" waaaay too frequently lately and i've got a bunch of tools that did One Thing, That One Time, and haven't been used since
|
# ? Apr 11, 2016 20:01 |
|
JawnV6 posted:btw, I'm on one end of the pendulum swing where I'm avoiding building tools. wondering if anyone else hits this point same and i also have a bunch of half-assed one-off tools that have hardcoded paths like /home/crip_eatin_bread/data/foobar.bin
|
# ? Apr 11, 2016 20:06 |
|
JawnV6 posted:btw, I'm on one end of the pendulum swing where I'm avoiding building tools. wondering if anyone else hits this point i've had the best luck when i look for existing tools first. there is often something that can be used to solve whatever problem i'm having
|
# ? Apr 11, 2016 20:08 |
|
git apparently lets you commit files with canonically equivalent pathnames. if i did the following on linux, would a subsequent `git checkout` just straight up fail on OS X, where all filenames are UTF-16 NFD strings, because OS X would think both pathnames refer to same file?pre:% git init % touch $'y\xc3\xb3spos.txt' # UTF-8 NFC (yóspos.txt) % touch $'yo\xcc\x81spos.txt' # UTF-8 NFD (yóspos.txt) % git commit -a -m 'pathname fuckery' pre:% git cat-file -p HEAD tree e05c87a903e9b87cda408fd12ff4a8b1d7767ffe <...> % git cat-file -p e05c87a903e9b87cda408fd12ff4a8b1d7767ffe 100644 blob ce013625030ba8dba906f756967f9e9ca394464a "yo\314\201spos.txt" 100644 blob cc628ccd10742baea8241c5924df992b5c019f71 "y\303\263spos.txt"
|
# ? Apr 11, 2016 20:23 |
|
JawnV6 posted:Open up notepad and quit pretending you need a fancier tool i already got an excel spreadsheet. the thing is, i do have to coordinate with a few other people (namely the DBA and the QA lead) and all of us sending 3,000 line spreadsheets to each other to look over and compare seems like it could be improved upon. also i am afraid of sending my test sheet out to just rot in someone's email forever (there have been offhand comments to suggest this has already happened). also i don't know a goddamn thing about automated tests except for making a pre-generated unit test in junit and i don't even remember how to do that offhand.
|
# ? Apr 11, 2016 21:38 |
|
what is it you do? write that down. you don't "do" a 3,000 line spreadsheet, you're running a program or doing some tasks with the program and each line is a test? each line is a transaction? idk, idc what generates this 3kline spreadsheet? is there any summary or subset of that that you could send to the dba/qa?
|
# ? Apr 11, 2016 21:45 |
|
think about what your test is actually doing write code to do that
|
# ? Apr 11, 2016 21:45 |
|
JawnV6 posted:what generates this 3kline spreadsheet? is there any summary or subset of that that you could send to the dba/qa? literally just typing out a script for each test ("do this, then do this"). i guess i could send just the list of test, but the script is needed by qa
|
# ? Apr 11, 2016 22:00 |
|
so you have 3,000 test cases that are specified for execution by humans these are run once a (week/day/release/month/moon) by hand and the results are put into a spreadsheet (guessing there's more than just a pass/fail column?) that is sent to the DBA & QA leads assuming that's an accurate description, going through all 3000 of those uses and documenting the result seems like the first step to automate. look into things like AutoIT (will capture your mouse & kb inputs, then let you replay them) or this SO for a more programmatic way
|
# ? Apr 11, 2016 22:05 |
|
JawnV6 posted:guessing there's more than just a pass/fail column? for the qa people to fill out? lol, no this is going to get run once per "release". "nah, i'm sure it all works fine" has been the only testing done before now no one knows what they are doing. internal legacy software is fun
|
# ? Apr 11, 2016 22:38 |
|
read a book specifically "working effectively with legacy code" by Michael C. Feathers
|
# ? Apr 11, 2016 22:40 |
|
i have read zero books about programming
|
# ? Apr 11, 2016 22:44 |
|
nothing wrong with that but if you come into the bad programming thread asking how to be a better programmer then i dont think being told to read a book is bad
|
# ? Apr 11, 2016 22:46 |
|
the only programming book I've read is "the art of unit testing" which I quite enjoyed
|
# ? Apr 11, 2016 22:51 |
|
Awia posted:nothing wrong with that it is not bad, but it is also not the only solution
|
# ? Apr 11, 2016 22:53 |
|
Bloody, show me the way
|
# ? Apr 11, 2016 23:01 |
|
compuserved posted:git apparently lets you commit files with canonically equivalent pathnames. if i did the following on linux, would a subsequent `git checkout` just straight up fail on OS X, where all filenames are UTF-16 NFD strings, because OS X would think both pathnames refer to same file? did Linus ever go back on his "all filesystems must store names bit-identically to what their APIs are given or they're broken and not worth supporting" idiocy? that was his position a few years ago, that färt and färt should both be allowed in a directory if they're encoded differently, and anything else is verboten and won't be supported by git or Linux
|
# ? Apr 11, 2016 23:02 |
|
eschaton posted:did Linus ever go back on his "all filesystems must store names bit-identically to what their APIs are given or they're broken and not worth supporting" idiocy? idk but ya that's pretty idiotic imo
|
# ? Apr 11, 2016 23:11 |
|
HoboMan posted:Bloody, show me the way what do you want to do in your career is probably a good question
|
# ? Apr 11, 2016 23:35 |
|
HoboMan posted:Bloody, show me the way very carefully
|
# ? Apr 11, 2016 23:36 |
|
i live an entirely isolated life at work and have improved my programmin skills by faffing around on company time and doing things like: reading this thread reading other yosthreads reading yosblogs (tef, mononcqc) reading nonyosblogs (danluu) faffing around with idiot projects (haskell most recently) managing my own workflow (jirad myself)
|
# ? Apr 11, 2016 23:38 |
|
I went lovely Python dev -> DevOps -> Continual Delivery Engineer and most of that was just paying attention to details, listening to people, loving around with code as much as possible and being the only person who wrote unit tests in a hiring code test
|
# ? Apr 11, 2016 23:47 |
|
eschaton posted:did Linus ever go back on his "all filesystems must store names bit-identically to what their APIs are given or they're broken and not worth supporting" idiocy? I thought Unix filenames were opaque sequences of bytes, so this would be a standpoint consistent with that
|
# ? Apr 12, 2016 00:20 |
|
ahmeni posted:I went lovely Python dev -> DevOps -> Continual Delivery Engineer and most of that was just paying attention to details, listening to people, loving around with code as much as possible and being the only person who wrote unit tests in a hiring code test so what the gently caress is DevOps?
|
# ? Apr 12, 2016 00:35 |
|
we make ur beep go boop
|
# ? Apr 12, 2016 00:38 |
|
hey fun c++03 fact: if you have a std::vector<Object*> v, then v[i]+1 compiles fine and without warning!
|
# ? Apr 12, 2016 00:40 |
|
Bloody posted:i live an entirely isolated life at work and have improved my programmin skills by faffing around on company time and doing things like: this is the first thing i looked at after google "tef blog": http://programmingisterrible.com/ i don't know if it's his, but it looks like it belongs here
|
# ? Apr 12, 2016 00:40 |
|
HoboMan posted:for the qa people to fill out? lol, no ok, it sounds like there's a legacy QA process and a new 3k one you're trying to spin up and my questions are alternately landing on either side of that divide so imma retreat to: 1) document current 2) improve from there
|
# ? Apr 12, 2016 00:42 |
|
HoboMan posted:so what the gently caress is DevOps?
|
# ? Apr 12, 2016 00:57 |
|
|
# ? May 25, 2024 14:13 |
|
eschaton posted:did Linus ever go back on his "all filesystems must store names bit-identically to what their APIs are given or they're broken and not worth supporting" idiocy? git config --global core.precomposeunicode true
|
# ? Apr 12, 2016 01:40 |