|
I type way too little code to have something to commit every other hour or so.
|
#
?
Jan 17, 2016 16:39
|
|
|
|
| # ? May 25, 2024 00:49 |
|
|
pointsofdata posted:I'm upset and I make excel plugins for a living. Did you know that the version of excel everyone uses has a 1.3gb address space limit? And that people are big fans of running large simulations in excel?
|
|
#
?
Jan 17, 2016 16:39
|
|
|
Valeyard posted:no, but i have learned recently that .xls documents (as apposed to .xlsx) have a row and column limit of 65536 old versions of excel (<= excel 95?) have only 16384 rows. found that out when a turkish data vendor sent us a file (in like 2013) that was mysteriously cut off, probably because they ran multiple crazy old versions and accidentally opened the file in the wrong one. pointsofdata posted:I'm upset and I make excel plugins for a living. Did you know that the version of excel everyone uses has a 1.3gb address space limit? And that people are big fans of running large simulations in excel? uh just parallelize your vba code across multiple excel instances. that way you'll actually use more than one processor too. it's trivial. ShimaTetsuo fucked around with this message at 16:47 on Jan 17, 2016 |
|
#
?
Jan 17, 2016 16:41
|
|
|
coffeetable posted:that was flippant of me, but you really should commit every.drat.thing you write, and you should commit often (like, at least once an hour). learn the shortcuts for your source control gui so you can make a checkpoint commit with a few second's effort. Well yeah, this is what I do normally but when doing tiny projects like this it's not really the first thing that comes to mind. Hilarious enough this is the second time this has happened to me though so maybe next time I'll remember (I won't)
|
|
#
?
Jan 17, 2016 17:03
|
|
|
is it worth using git if I'm the only one that is ever going to use my programmes?
|
|
#
?
Jan 17, 2016 17:08
|
|
|
yeah
|
|
#
?
Jan 17, 2016 17:11
|
|
|
vodkat posted:is it worth using git if I'm the only one that is ever going to use my programmes? (with a GUI and github) git backs stuff up online, lets you switch between computers easily, lets you revert. i don't use branches and stuff when working on hobby projects but i probably should, just to separate concerns and stop me jumping between 3 half done features
|
|
#
?
Jan 17, 2016 17:58
|
|
|
I'm still constantly amazed when I write threaded code and it appears to work as intended can't wait for something to finally fail in production
|
|
#
?
Jan 17, 2016 18:25
|
|
|
vodkat posted:is it worth using git if I'm the only one that is ever going to use my programmes? that's nearly the best feature of git. the per-project overhead is almost zero because you don't have to fart around with a vcs server or anything if you don't want
|
|
#
?
Jan 17, 2016 18:37
|
|
|
gonadic io posted:(with a GUI and github) git backs stuff up online, lets you switch between computers easily, lets you revert. i don't use branches and stuff when working on hobby projects but i probably should, just to separate concerns and stop me jumping between 3 half done features i tend to git yolo on personal projects but at least that bad habit doesnt seem to be loving up my at work workflow
|
|
#
?
Jan 17, 2016 18:38
|
|
|
i use branches much more heavily for personal projects than for real work things, actually. for work 90% of my branches are just things to be code reviewed before merging into master, but for personal projects i often start implementing something, hit a bug i need to fix first, realize i need to do some refactoring for that, and then suddenly i have eight live branches all with unrelated half-finished changes
|
|
#
?
Jan 17, 2016 18:39
|
|
|
Plorkyeran posted:i use branches much more heavily for personal projects than for real work things, actually. for work 90% of my branches are just things to be code reviewed before merging into master, but for personal projects i often start implementing something, hit a bug i need to fix first, realize i need to do some refactoring for that, and then suddenly i have eight live branches all with unrelated half-finished changes i just do this all in master then find out that i need to revert one of those changes but not the rest. but then i've always been a glutton for punishment sooo
|
|
#
?
Jan 17, 2016 19:00
|
|
|
i commit all the time. like if i change one line of code and it changes behavior or fixes something then i commit just for that change. i dont understand why you'd wait hours between commits unless you werent actually typing anything.
|
|
#
?
Jan 17, 2016 19:16
|
|
|
coffeetable posted:i want to bitch about what a pain it is to get a program i developed on a machine w/ 32GB memory running on a 8GB production server, but i think the embedded devs itt might get upset why cant you run in production on a 32gb machine also you should be able to impose memory limits in like every framework ever
|
|
#
?
Jan 17, 2016 19:38
|
|
|
fart simpson posted:i commit all the time. like if i change one line of code and it changes behavior or fixes something then i commit just for that change. i dont understand why you'd wait hours between commits unless you werent actually typing anything. At work I commit rarely (in the feature branch) , mostly only when I finish big stages - the first draft, the complete implementation, then the tests. And then at the end of each day. I don't like having commits of non-compiling code. At home I commit whenever get it to do something new without crashing on launch. If I ever want to actually sell this game at some point I'm going to need a better testing cycle (I have unit tests around some of the core algs and data structures)
|
|
#
?
Jan 17, 2016 19:45
|
|
|
Also my team had a good laugh over my personal commit messages when they were hiring me (and then checked that my ones at work we're at least vaguely professional or even slightly informative)
|
|
#
?
Jan 17, 2016 19:49
|
|
|
i commit garbage constantly until i get to a point where the work done represents something atomic and then i rebase that into a single commit or a group of commits that make sense logically.
|
|
#
?
Jan 17, 2016 20:02
|
|
|
~Coxy posted:i would be interested in trying a source control system or plugin that would automatically give you other people's changes as they get checked in
|
|
#
?
Jan 17, 2016 20:09
|
|
|
vodkat posted:is it worth using git if I'm the only one that is ever going to use my programmes? anything that is text that might benefit from version control is a candidate basically
|
|
#
?
Jan 17, 2016 20:11
|
|
|
vodkat posted:is it worth using git if I'm the only one that is ever going to use my programmes? yes, absolutely
|
|
#
?
Jan 17, 2016 20:32
|
|
|
gonadic io posted:At work I commit rarely (in the feature branch) , mostly only when I finish big stages - the first draft, the complete implementation, then the tests. And then at the end of each day. I don't like having commits of non-compiling code. people who do this and the other "linear history" git trope are so weird; why does every commit have to be some elegant encapsulation of a specific concept? i'm one of the biggest perfectionists I know of and my git log looks like a terrible programmers stream-of-consciousness diary interspersed with machine-gunned WIPs because git is for protecting your work, not building some kind of software dev fairy tale
|
|
#
?
Jan 17, 2016 20:35
|
|
|
Corla Plankun posted:people who do this and the other "linear history" git trope are so weird; why does every commit have to be some elegant encapsulation of a specific concept? I think there are two schools of thought that are at odds: The first is that your commits should be encapsulated and they should always compile/pass tests/work/etc. The other is that you should commit everything you do. Those two ideas are fundamentally at odds. I think the confusion is that the work you submit to others should be well encapsulated. But that doesn't mean your awful stream of consciousness work shouldn't be committed in a working branch. I am just completely unable to do "work in progress" stuff that is well encapsulated every step of the way. My brain does not work like that. I have broken bullshit all over the place while I'm working on a change. I think some people are able to take a more organized approach, but that simply doesn't work for me.
|
|
#
?
Jan 17, 2016 20:42
|
|
|
oh don't get me wrong, my partial commits only compile due to judicious use of "throw new NotYetImplementedException" or whatever your language of choice - "???" for scala and "undefined" for haskell
|
|
#
?
Jan 17, 2016 21:08
|
|
|
i commit stream of consciousness and then clean it up when i feel it's in a good state that represents an atomic set of work "git rebase --interactive" makes this trivial it's also a good practice to separate refactorings/code clean-up from actual functional code changes and the above command and a slight amount of discipline makes this easy as well
|
|
#
?
Jan 17, 2016 21:08
|
|
|
my method that i highly highly recommend trying is to make a copy of my work branch and then do `git reset --mixed master` (or whatever branch your work is originally based from). This basically just unstages all changes in the branch so far. Then I use a graphical tool like git cola (despite using console stuff for everything else, this is one thing where I've found a graphical tool to be really superior for some reason) that lets me go through the changes and stage them line by line into a group of logical commits. At the end I'll have 3 or 4 commits that maybe make sense together. That way, when someone is reviewing the work, I can tell them to look at the individual commits to get a better sense of what the changes were about. This isn't something I do for all of my work, but it's worth the effort when I'm doing a bunch of work and I feel bad for the reviewers. Also I do it because my wip branches are particularly bad (like, sometimes over 100 'wip' commits) and rebase really doesn't cut it for me. This approach lets me get a good holistic view of the work I've done and usually I end up catching obvious problems. If you are better at organizing your work it's probably overkill. DONT THREAD ON ME fucked around with this message at 21:24 on Jan 17, 2016 |
|
#
?
Jan 17, 2016 21:16
|
|
|
i generally commit a bunch to my local git and then push it when it's actually compilable
|
|
#
?
Jan 17, 2016 21:23
|
|
|
my branches can get messy but anything that's going to be shared by other people gets rebased and squashed down to discrete feature commits. also if you commit garbage you break git bisect which is like it's best feature
|
|
#
?
Jan 17, 2016 21:34
|
|
|
the talent deficit posted:my branches can get messy but anything that's going to be shared by other people gets rebased and squashed down to discrete feature commits. also if you commit garbage you break git bisect which is like it's best feature you almost never need 'git bisect', but when you need it, you really need it having a clean commit log is important for those once-a-year "oh poo poo when did it break" moments
|
|
#
?
Jan 17, 2016 22:01
|
|
|
Whenever I'm contributing to something that other people are also working on, I commit like crazy locally and then rebase it to make it neat before making a pull request.
|
|
#
?
Jan 17, 2016 22:15
|
|
|
whenever i see someone teaching a git newbie to clean up their commits with squash / rebase, some days later the same person will be back to help them figure where their changes disappeared to
|
|
#
?
Jan 17, 2016 22:34
|
|
|
The commits in dev are almost entirely 'merged branch feature/jira-id-butt into develop' and in master are merging develop into master. I don't see the point (and have never been asked to) rebase my local commits
gonadic io fucked around with this message at 23:26 on Jan 17, 2016 |
|
#
?
Jan 17, 2016 22:43
|
|
|
can you squash commits in mercurial without an extension
|
|
#
?
Jan 17, 2016 22:53
|
|
|
I have two github accounts A and B. A has access to private repo R, B does not. I download and install Sourcetree (because Github for Windows refuses to work) and login with user A, and clone R. i make two commits to R, both have now mysteriously appeared as user B. it then lets me push the commits, done as user B, to repo R so B never had access to R, and B was not the account logged into Sourcetree, and the end result is it pushed to the repo anyway. From account B i couldnt see the repo R or my contributions, but if I logged in with A I could see that B has made commits to it cool
|
|
#
?
Jan 18, 2016 00:24
|
|
|
why do you think GitHub authenticates based on the username on the commits
|
|
#
?
Jan 18, 2016 00:47
|
|
|
B can't see anything about the repo because it does not have access to view information about the remote repository. B can make commits toward the local repository because there's no security on your workstation. names on commits are not secured in anyway, are super arbitrary, and people have used that to gently caress with notable people. you should check your gitconfig. A can push commits to the remote, but because of the above they can have arbitrary metadata.
|
|
#
?
Jan 18, 2016 00:54
|
|
|
FamDav posted:B can't see anything about the repo because it does not have access to view information about the remote repository. i guess, i forgot you can put in random names
|
|
#
?
Jan 18, 2016 01:01
|
|
|
hello thread, im goign to transition from a terrible java dev to a terrible c# dev with a $13K pay rise  believe in yourselfs!!
|
|
#
?
Jan 18, 2016 01:06
|
|
|
bobbilljim posted:hello thread, im goign to transition from a terrible java dev to a terrible c# dev with a $13K pay rise congratulations on successfully moving from one of the good languages to the other!
|
|
#
?
Jan 18, 2016 01:12
|
|
|
dont worry, i'll still be using javascript also 
|
|
#
?
Jan 18, 2016 01:26
|
|
|
|
| # ? May 25, 2024 00:49 |
|
|
i moved from java to python to node to go to node/python and angular the things i do for pay raises 
|
|
#
?
Jan 18, 2016 02:12
|
|