Package Details: bup-git 1:0.32.r174.g45d1fd23-1

Git Clone URL: (read-only, click to copy)
Package Base: bup-git
Description: Efficient file backup system based on the git packfile format
Upstream URL:
Keywords: backup dev git packfile system
Licenses: GPL
Conflicts: bup
Provides: bup
Submitter: mfwitten
Maintainer: edh
Last Packager: edh
Votes: 22
Popularity: 0.000000
First Submitted: 2011-04-20 17:55 (UTC)
Last Updated: 2021-08-17 13:57 (UTC)

Required by (4)

Sources (1)

Latest Comments

edh commented on 2021-08-17 13:57 (UTC)

@chowbok done.

chowbok commented on 2021-08-17 13:30 (UTC)

Hi! Would you mind adding "aarch64" to the list of supported architectures? It works fine on there.

edh commented on 2021-03-02 22:10 (UTC)

@mederel An additional commit would be completely sufficient but yes. IMHO this change does not merit introducing an epoch.

If you are annoyed by your package manager reminding you to update, just forcefully rebuild the package and you should be fine.

mederel commented on 2021-03-02 22:01 (UTC)

Hi @edh, I take it then that we await the next upstream release, and things should go back to normal, aren't they? Regards.

edh commented on 2021-02-16 12:45 (UTC)

@mederel and @IslandC0der Apparently the problem really was at my end. I nuked the repository and cloned again. Now I get a checksum that is one character shorter.

IslandC0der commented on 2021-02-16 12:35 (UTC)

Hmmm, I think @menderel is right; I'm getting the same issue just by running makepkg -do, no AUR helpers. The pkgver gets updated to 1:0.32.r0.g3b052c6-1

If you visit the github page, the latest commit is indeed 3b052c6 (note the missing '3' at the end), but if you click on it, you can see the full id, which is 3b052c63c93a95193447945296e7ec5db97f6b6e (there is a '3' right after). So now the question is: why is your pkgver() extracting one extra character, @edh?

mederel commented on 2021-02-15 20:36 (UTC)

@IslandC0der and @edh, thank you for your answers!!! And my apologies for a (2-week-long awaited!! :-( ) late answer.

Looking at pkgver I see the following command which I ran on a copy of the repository:

$ git describe --long 2>/dev/null

So here the 3 is missing. Just need to find why.

IslandC0der commented on 2021-01-31 22:14 (UTC)

@mederel, I think @edh is right

Are you using yay? If so, have you tried doing a clean build?

edh commented on 2021-01-31 18:01 (UTC)


The pkgver is computed correctly and is equal to the one here at the AUR. I can further compile the package just fine and thereby reproduce the exact same pkgver. This just might be a bug in your AUR helper and not in the PKGBUILD.

mederel commented on 2021-01-31 17:35 (UTC)

Every system update I do I see this AUR update proposed:

2  aur/bup-git    1:0.32.r0.g3b052c6-1 -> 1:0.32.r0.g3b052c63-1

It has to do with the version that is computed in the PKGBUILD I believe. Tried to change it accordingly in the file but it did nothing the update continue to come. Any idea where this could be comming from?

IslandC0der commented on 2020-08-23 11:04 (UTC) (edited on 2020-08-23 11:04 (UTC) by IslandC0der)

Ohh, nevermind, I was missing some other dependencies apparently.

Anyways, thanks for letting me know!

IslandC0der commented on 2020-08-23 10:47 (UTC)

Hmm, have you tried running bup fuse with the new dependencies instead of the python2 ones? I get a error: cannot find the python "fuse" module; please install it unfortunately :/

edh commented on 2020-08-23 09:51 (UTC)

Bup has finally be switched to python3.

@IslandC0der Thanks for uploading the old python2 dependencies in the meantime.

IslandC0der commented on 2020-07-15 20:17 (UTC)

Ohh, I see. My bad, I'll submit python2-fuse as well then.

edh commented on 2020-07-15 20:14 (UTC)

bup fuse doesn't work if python2-fuse is not installed. check succeeding actually makes sense as python2-fuse is probably not at the core of bup and may even be considered optional.

IslandC0der commented on 2020-07-15 20:09 (UTC)

Hmm, have you run into any issues? By replacing the package, I could execute the check function without problems.

edh commented on 2020-07-15 19:56 (UTC) (edited on 2020-07-15 19:59 (UTC) by edh)

@IslandC0der Thanks! ~~I'll do that then.~~

EDIT: python2-fuse can not be replaced by python-fuse!

IslandC0der commented on 2020-07-15 19:29 (UTC)

Nice, that looks promising, let's hope it comes out soon. In the meantime, I've submitted python2-pyxattr to the AUR, and you can replace the old python2-fuse dependency with python-fuse, and it works just the same.

edh commented on 2020-07-11 15:03 (UTC) (edited on 2020-07-11 15:06 (UTC) by edh)

@IslandC0der Thanks for the reminder. TBH I am thinking about dropping the package as I am not willing to support a package that uses an unsupported programming language. python2 should be dropped and be replaced by python3. IMHO just picking up the peaces and taking care of a bunch of deprecated python2 packages does not sound like a solution.

Let's hope that a new release lands soon [1].


IslandC0der commented on 2020-07-11 14:38 (UTC)

python2-fuse and python2-pyxattr were dropped from the repos.

edh commented on 2016-05-08 14:27 (UTC)

@bartus Thanks for the hint. Apparently the documentation [1] about the PREFIX is wrong then. Concerning your previous comment, which you deleted, I am very interested in any suggestion about improving the PKGBUILD. However using roughly the same PKGBUILD as the bup package is unfortunately not possible since the build function would error out. [1]

bartus commented on 2016-05-08 10:06 (UTC) (edited on 2016-05-08 11:12 (UTC) by bartus)

Add PREFIX='/usr/' to package make invocation to move bup form /usr/local to /usr.

edh commented on 2015-06-08 10:44 (UTC)

Hi, I am the new maintainer and will update the package from now on. The mentioned errors should be fixed and the package should work just fine (tested on two computers).

esh commented on 2014-12-17 14:08 (UTC)

The error (/usr/bin/makepkg: eval: line 2225: unexpected EOF while looking for matching `)') is due to the comment inside the options=() array. If you delete the comment or move it down one line (after the ')'), it works just fine.

mfwitten commented on 2014-05-12 01:55 (UTC)

Arch Linux is increasingly a pile of poo.

bfr commented on 2014-04-28 22:06 (UTC)

makepkg seems to have problems parsing/ignoring the parentheses in the comment of the PGKBUILD-options. /usr/bin/makepkg: eval: line 2225: unexpected EOF while looking for matching `)'

mfwitten commented on 2014-04-19 19:31 (UTC)

Updated. Thanks

bfr commented on 2014-04-18 15:43 (UTC)

Could you please add !makeflags to the PGKBUILD-options? I have MAKEFLAGS="-j8" in /etc/makepkg.conf and 'make test' fails. After a lot of debugging I found that it is a race condition with test -e t/tmp || mkdir t/tmp in line 84 and 88 in the Makefile. 'make test' fails with mkdir: cannot create directory ‘t/tmp’: File exists

mfwitten commented on 2014-02-18 15:58 (UTC)

Thanks, rakoo.

rakoo commented on 2014-02-16 03:43 (UTC)

There's a sed missing in t/ Here's a diff for it: --- PKGBUILD.orig 2014-02-16 04:42:26.774651668 +0100 +++ PKGBUILD 2014-02-16 04:39:10.545558824 +0100 @@ -97,7 +97,7 @@ sed -i -e 's/PYTHON=python\b/PYTHON=python2/' -e '/docs-available/d' Makefile - sed -i -e 's/python\b -c/python2 -c/' t/{lib,test-meta}.sh + sed -i -e 's/python\b -c/python2 -c/' t/{lib,test-meta,test-ls}.sh fi }

bfr commented on 2013-12-14 02:24 (UTC)

I can confirm it's working now. Don't know what went wrong the last time, I can't reproduce it. Sorry to bother you. :-)

mfwitten commented on 2013-12-14 01:47 (UTC)

Commit 5a7fd007e89bb343ab2024d9a87d54bbee997bbf seems to build/test/install just fine via makepkg. Does your current working directory path use a symlink? There have at least been historical bugs related to that, and those bugs may still exist.

bfr commented on 2013-12-10 23:09 (UTC)

make test works from cloned git repo, but not when called from makepkg. Does anyone know why?

mfwitten commented on 2013-11-05 20:18 (UTC)

Thanks, LongHairedHacker. I just added `t/' directly to the relevant command line, as I'd rather be conservative about such things.

LongHairedHacker commented on 2013-11-04 21:34 (UTC)

Hey just short notice your sed commands miss one occurrence of python. There is one in t/ So sed -i -e 's/python\b -c/python2 -c/' t/*.sh instead of sed -i -e 's/python\b -c/python2 -c/' t/ should fix that.

mfwitten commented on 2013-09-02 04:24 (UTC)

This one updates how git is accessed, as per `man PKGBUILD'. Sorry for any trouble this may cause people.

commented on 2013-08-30 11:47 (UTC)

Hmmm. so make test doesn't work on tmpfs it seems, so yaourt will fail to build it. Hand rolling works though. I've sent a msg to the mailing list, might get a change to look at it over the weekend.

commented on 2013-08-30 09:31 (UTC)

Broken on current git, needs the following change: --- old 2013-08-30 10:25:34.240928254 +0100 +++ new 2013-08-30 10:26:14.419835249 +0100 @@ -100,6 +100,7 @@ sed -i -e 's/acLookFor python\b/acLookFor python2/' \ -e 's/MF_PATH_INCLUDE PYTHON python\b/MF_PATH_INCLUDE PYTHON python2/' \ + -e 's/bup_find_prog python/bup_find_prog python2/' \ config/configure sed -i -e 's/PYTHON=python\b/PYTHON=python2/' -e '/docs-available/d' Makefile Tests are also failing, but that is probably a different issue with my filesystem.

commented on 2013-04-26 09:58 (UTC)

Yoa are right, found it: Sorry for the trouble.

mfwitten commented on 2013-04-25 04:12 (UTC)

My first inclination is that this is a problem with the Python package itself.

commented on 2013-04-24 20:43 (UTC)

Hello, this PKGBUILD seems broken. Tried it on multiple Arch Systems. In file included from /usr/include/python2.7/Python.h:126:0, from _helpers.c:6: /usr/include/python2.7/modsupport.h:27:1: Fehler: »_PyArg_ParseTuple_SizeT« ist ein nicht erkannter Formatfunktionstyp [-Werror=format=] PyAPI_FUNC(int) PyArg_ParseTuple(PyObject *, const char *, ...) Py_FORMAT_PARSETUPLE(PyArg_ParseTuple, 2, 3); ^ Greetings maralorn

eliasson commented on 2013-02-01 17:33 (UTC)

Please add rsync to the makedepends. Without it the tests won't run properly.

mfwitten commented on 2013-01-07 03:58 (UTC)

Thanks, Lucky. I've updated it the repo URI; sorry for the delay.

Lucky commented on 2013-01-04 16:02 (UTC)

hi, please use this repo because its now the offical one. see description on old repo ( "Please switch to using the repo instead!"

mfwitten commented on 2012-09-19 15:22 (UTC)

That is a known bug caused by developments in `git'; see here for a patch that should be making its way into the bup repo:

temoto commented on 2012-09-18 20:59 (UTC)

pacaur -S bup-git test failure

commented on 2012-06-17 18:04 (UTC)

the package also doesn't install on a freshly installed system (when "python" -- version 3 -- is not installed). you probably need to perform the sed from "python" to "python2" on config/configure

commented on 2012-06-17 17:58 (UTC)

bup calls out to git on some occasions (when listing commits for example) when running, so you probably want to add git to the "depends" list.

ever commented on 2012-06-01 15:36 (UTC)

Please change the source url to a http one. Else people behind a firewall (but with a working http proxy) cannot build it.

alperkanat commented on 2012-05-20 20:13 (UTC)

can't compile due to a test failure :(

commented on 2012-05-13 06:28 (UTC)

Indeed it is. Thanks a lot!

mfwitten commented on 2012-05-13 02:39 (UTC)

Thanks, aghitza; it's fixed.

commented on 2012-05-12 23:33 (UTC)

I'm getting "Dependency `pyxattr' of `bup-git' does not exist. I think pyxattr has been renamed python2-pyxattr. Would you mind updating the PKGBUILD?

mfwitten commented on 2011-04-21 02:21 (UTC)

I pretty much use soley `vim'; I find that while it makes medium tasks easy, it makes hard tasks as annoying as possible (it definitely feels like the pieces were shoehorned together and bound with chewing gum and twine by many different people over the course of very many years). :-) Why do you ask?

L42y commented on 2011-04-21 01:06 (UTC)

PKGBUILD file is interesting, what's your editor?