|
cool av posted:sure...but there's value to be found in the "garbage" (the commit history as it really happened). is it too much to ask to be able to see it when I want to, and filter it out when I don't? There absolutely isn't
|
# ? Feb 9, 2021 07:33 |
|
|
# ? Jun 8, 2024 18:28 |
|
cool av posted:i'm mostly just saying git should have a way of tracking which commits are intended to be consumed by which tools (eg. bisect) or pipelines, and not lose that information when you merge and force you to rely on a separate database to track it. What? Wat? WAT?
|
# ? Feb 9, 2021 07:34 |
|
you see, git provides room for the holy wars and does not get involved in them itself. dvcs holy war generating machine, really
|
# ? Feb 9, 2021 07:35 |
|
Of course git was built for source control, what the gently caress would it be built for? The history is literally that linux kernel needed new source control sw, and git was created for that. And I assume in your weird world, only merge commits are intended for ci, bisecting, etc, and guess what, you can do that. Our resident psyduck has even already said how, you go through first parents of merge commits.
|
# ? Feb 9, 2021 07:40 |
|
The unsung 13th labor of hercules, caring about the first parent merge commit option for bisects.
|
# ? Feb 9, 2021 08:47 |
|
my fav git story is torvalds mentioning in passing early on (might've been in this 2007 tech talk? idk) that a lot of git's design decisions were based on looking at whatever svn did and doing exactly the opposite branches in the source tree? nah lets put source trees in branches sequential commits? screw that lets have random hashes referencing objects by path? nope gonna just hash that too resolving merge conflicts by hand on the server? just force the developer to deal with it before they push etc
|
# ? Feb 9, 2021 12:22 |
|
Look, again, workflows are all based on need. There's a lot of good rationale for everything here. Bisect, squash, do whatever, just don't be a loving idiot, and make sure you use Emacs for merging. You'll gently caress it all up otherwise, just like the rest of your life choices.
|
# ? Feb 9, 2021 15:09 |
|
Progressive JPEG posted:my fav git story is torvalds mentioning in passing early on (might've been in this 2007 tech talk? idk) that a lot of git's design decisions were based on looking at whatever svn did and doing exactly the opposite every single one of those decisions was cool and good sothat was a great decisionn heuristic
|
# ? Feb 9, 2021 18:11 |
|
Nomnom Cookie posted:no, there isn't value in 50 commits with log messages in varying combinations of "fix", "gently caress", "please", "try", and "this time" speak for yourself! they have sentimental value. but seriously even if you personally never needed it, it is useful sometimes. for instance if other branches were merged into your branch at various times. Xarn posted:Of course git was built for source control, what the gently caress would it be built for? The history is literally that linux kernel needed new source control sw, and git was created for that. first-parent is a ridiculous hack that sort of works if you're careful (a common theme in git)
|
# ? Feb 9, 2021 19:31 |
|
cool av posted:but seriously even if you personally never needed it, it is useful sometimes. for instance if other branches were merged into your branch at various times. dont do that. you have main. branch off that, then squash merge. thats it. if you feel a need to get fancier then your branches are living too long. stop doing that
|
# ? Feb 9, 2021 20:10 |
|
"I could not replicate the issue because the same session is used in all tabs" the issue literally says "if you open multiple tabs the same session is used and one tab overwrites the object in the other" I mean ffs
|
# ? Feb 9, 2021 20:12 |
|
i think devs naturally become really good at telling users they don't want what they want, for reasons, when the real reason is what they want would require a significant amount of rework. heck, we're proud of it a lot of the time and probably right a lot of the time too.
|
# ? Feb 9, 2021 20:35 |
|
"you can have this for a nickel or that for 700,000 dollars"
|
# ? Feb 10, 2021 00:32 |
|
cool av posted:i think devs naturally become really good at telling users they don't want what they want, for reasons, when the real reason is what they want would require a significant amount of rework. On the flip side of this users/"product guys" often don't really understand what they are asking for or how hard a problem is to actually solve completely. I've made search systems before where people get upset at some aspect of the order of returned and say that I should have it sort by "best match." it's only when you can get them to try and describe what "best match" means in general and not just how they wanted this particular query to come back that they realize what they are asking for is for me to make the computer read minds.
|
# ? Feb 10, 2021 01:18 |
|
Zaxxon posted:how they wanted this particular query to come back that they realize what they are asking for is for me to make the computer read minds. "aight we'll give it a ton of tunables" "now it's too complicated"
|
# ? Feb 10, 2021 03:59 |
|
i've used a ton of them, but opendiff is still the best diff tool i've used for merge conflicts
|
# ? Feb 10, 2021 04:13 |
|
Achmed Jones posted:i've used a ton of them, but opendiff is still the best diff tool i've used for merge conflicts whats good about opendiff? I really like pycharms as well. Merge all non conflicting in one click and usually 10 seconds to deconflict most issues.
|
# ? Feb 10, 2021 06:23 |
|
git pull origin master -Xours && git push --force bing bong so simple.
|
# ? Feb 10, 2021 06:24 |
|
My work has us use the global gitignore file for os-specific files. So instead of adding desktop.ini or .ds_store to the repo's .gitignore once and being done with it, every person who works on it has to do the same thing
|
# ? Feb 10, 2021 07:11 |
|
Do you at least have a lot of repos?
|
# ? Feb 10, 2021 08:06 |
|
cool av posted:i think devs naturally become really good at telling users they don't want what they want, for reasons, when the real reason is what they want would require a significant amount of rework. I spend a lot of time doing this because I went sideways into the dev side from being a
|
# ? Feb 10, 2021 12:41 |
|
brand engager posted:My work has us use the global gitignore file for os-specific files. So instead of adding desktop.ini or .ds_store to the repo's .gitignore once and being done with it, every person who works on it has to do the same thing that's the right way to do it otherwise every single repo's .gitignore has to have the os files plus whatever crap each person's editor is putting into the project directory
|
# ? Feb 10, 2021 13:32 |
|
my homie dhall posted:that's the right way to do it it's technically right but practically impossible as the number of people involved goes beyond, like, 2. see also: tabs vs spaces do whatever you gotta do to keep crap out. nothing cares about a repo gitignore that's ten lines longer than theoretically necessary
|
# ? Feb 10, 2021 15:38 |
|
like I'm happy to set up a tools-appropriate global gitignore for myself, but I wouldn't even bother mentioning the existence of that feature to teammates, let alone insist that they use it
|
# ? Feb 10, 2021 15:39 |
|
take the passive-aggressive option of setting up a precommit hook that rejects commits containing those files and prints a link to documentation on setting up a global gitignore
|
# ? Feb 10, 2021 15:58 |
|
Jabor posted:take the passive-aggressive option of setting up a precommit hook that rejects commits containing those files and prints a link to documentation on setting up a global gitignore
|
# ? Feb 10, 2021 16:21 |
|
Jabor posted:take the passive-aggressive option of setting up a precommit hook that rejects commits containing those files and prints a link to documentation on setting up a global gitignore let's add "precommit hooks are not ci" to the list alongside "mirrors are not backups" and "javadocs are not documentation"
|
# ? Feb 10, 2021 17:39 |
|
Jabor posted:take the passive-aggressive option of setting up a precommit hook that rejects commits containing those files and prints a link to documentation on setting up a global gitignore nah take the aggressive option, build fails CI, gently caress you have a nit to pick? right into CI it goes. if you can't automate it, shut the gently caress up about it
|
# ? Feb 10, 2021 17:43 |
|
Sagacity posted:how will you distribute that hook though we have a few repos where the build script installs git hooks. it works about as well as a local check can.
|
# ? Feb 10, 2021 22:18 |
|
don't use git hooks, run ci checks as part of your pr process instead. i want to be able to make as many trash commits as i want on my private branch, and also commit hooks tend to be slow as poo poo and i really don't want to have to wait for a whole bunch of them to run in sequence when i'm doing an interactive rebase or something like that. the pr workflow should also only permit rebase merges, so that the ci checks test what actually gets merged
|
# ? Feb 10, 2021 22:39 |
|
As long as you run CI against merge commits, you don't have to limit yourself to rebase merges tho.
|
# ? Feb 10, 2021 22:53 |
|
|
# ? Feb 11, 2021 20:44 |
|
lmao
|
# ? Feb 11, 2021 23:18 |
|
if you bulled that bullshit here it would be an immediate sentencing of your hands being trampled by the company brontosaurus
|
# ? Feb 11, 2021 23:44 |
|
what a champ. reminds me, how good is the state of the art in determining whether 2 unicode strings "look the same to a person" these days?
|
# ? Feb 12, 2021 02:41 |
|
welp my dumb python / flask / uwsgi app isnt doing 100% cpu because of the one thread i had in there, as it still happens after we removed the route that could trigger it and it doesnt appear to be actual usage, cause it has jumped up both when under some live testing, and at 4 am when nobody was using it gonna have to use actual debugging i guess ughhhh
|
# ? Feb 12, 2021 08:08 |
|
now trying with one multithreaded uwsgi process and if that doesnt work, with several single-threaded uwsgi processes
|
# ? Feb 12, 2021 12:51 |
|
Legit question, can you generate a valgrind profile on Python? I've never actually looked into it.
|
# ? Feb 12, 2021 14:31 |
|
i have no idea how to debug python aside from lots of logging would like a good primer on it tbh. im pretty good with lldb and breakpoints and stuff, but python is a black box to me
|
# ? Feb 12, 2021 16:44 |
|
|
# ? Jun 8, 2024 18:28 |
|
https://docs.python.org/3/library/pdb.html ?
|
# ? Feb 12, 2021 17:12 |