Git 2.21 Released With Performance Improvements, Human Date Option

[phoronix]

Phoronix: Git 2.21 Released With Performance Improvements, Human Date Option

Git maintainer Junio Hamano has just announced the release of Git v2.21.0 with more than 500 commits since the previous release from more than six dozen developers...

http://www.phoronix.com/scan.php?page=news_item&px=Git-2.21-Released

[programmerjake]

the github blog link is broken. it's at https://github.blog/2019-02-24-highl...from-git-2-21/ rather than https://github.blog/2019-02-23-highl...from-git-2-21/

[Michael]

the github blog link is broken. it's at https://github.blog/2019-02-24-highl...from-git-2-21/ rather than https://github.blog/2019-02-23-highl...from-git-2-21/

Thanks, looks like they changed the URL.

[Danny3]

I wonder when will they fix the the most annoying problem of them all, Git being blind to empty directory ?

[nanonyme]

I wonder when will they fix the the most annoying problem of them all, Git being blind to empty directory ?

Why is this a problem?

[Danny3]

Why is this a problem?

Why do you not see it as a problem?
Git is a "changes tracking software", but still, it is completely blind to some changes.
Just create a new directory and leave it empty, the see what git status shows.
How is that not a problem ?
A lot of people use git (myself included) besides tracking changes to also sync changes. I mean also for the transfer of items (directories and files) instead of the old way with FTP.
Guess what happens with empty directories. They are silently skipped from the transfer because Git is blind and could not even see them in the first place.
Now just image that some of the directories you created because you know that some program or script will write something inside them later. Like the log directories.
After you push | pull | whatever you expect that git has done a complete sync, but actually it didn't and you program might show error or fails because it cannot write some log file to a not existant directory on your public server.
There are workarounds with putting some files into those folders, but it looks like crap if you ask me.

[Guest]

After you push | pull | whatever you expect that git has done a complete sync, but actually it didn't and you program might show error or fails because it cannot write some log file to a not existant directory on your public server.

Normally, people create .gitkeep files to keep track of directories which should be considered as empty by git. When it comes to programs failing to run because they don't have some directory for log files, the best approach is to ensure the directory exists by creating it if it doesn't. It's not that complex to pull off, and it makes your software much more robust.

[Danny3]

Normally, people create .gitkeep files to keep track of directories which should be considered as empty by git. When it comes to programs failing to run because they don't have some directory for log files, the best approach is to ensure the directory exists by creating it if it doesn't. It's not that complex to pull off, and it makes your software much more robust.

Yes, I know the workaround fix to help git to see it, but still I don't like because I have useless files for me that just pollutes my vision of empty directories.
And not only that it pollutes my vision and decrease a bit my productivity, but I think it will also give false numbers on the total count of files in some directory properties windows, if you need that number for something like for some comparison.
In any case I don't like and never will workarounds when the fix could be done directly into the core of the software.
Git is still an actively developed software, so it could be fixed, if they wanted, instead of leaving us the developers fill our filesystems with nonsensical .gitkeep files.
Anyway, with the rest you are right, the software could be more robust, to check everything before it's trying to to an action, but this means more development time, and not always you can afford it. It would've been nice if we could though.

[emblemparade]

I wish I had a human date option. Robots can't love you back.

[starshipeleven]

Why do you not see it as a problem?
Git is a "changes tracking software", but still, it is completely blind to some changes.

To be fair, VCS software are mostly designed to work on files, usually text files.

While some like SVN do store empty folders too, other VCS like Mercurial don't detect empty directories either.

I suggest to not hold your breath and either add the feature yourself or switch to SVN.