git-annex.git
4 years agoremove
Joey Hess [Tue, 19 Apr 2022 16:14:27 +0000 (12:14 -0400)]
remove

4 years agocomment
Joey Hess [Tue, 19 Apr 2022 16:14:12 +0000 (12:14 -0400)]
comment

4 years agocomment
Joey Hess [Tue, 19 Apr 2022 16:08:12 +0000 (12:08 -0400)]
comment

4 years agomulticast: Support uftp 5.0 by switching from aes256-cbc to aes256-gcm
Joey Hess [Tue, 19 Apr 2022 16:02:10 +0000 (12:02 -0400)]
multicast: Support uftp 5.0 by switching from aes256-cbc to aes256-gcm

aes256-gcm is supported by both 4.x and 5.x, while 5.x dropped aes256-cbc.

Sponsored-by: Graham Spencer on Patreon
4 years agocomment
Joey Hess [Tue, 19 Apr 2022 15:56:32 +0000 (11:56 -0400)]
comment

4 years agoAdded a comment
yarikoptic [Tue, 19 Apr 2022 14:40:18 +0000 (14:40 +0000)]
Added a comment

4 years agoreport on cryto test FAIL on NFS
yarikoptic [Tue, 19 Apr 2022 14:37:43 +0000 (14:37 +0000)]
report on cryto test FAIL on NFS

4 years ago(no commit message)
lh [Tue, 19 Apr 2022 01:50:23 +0000 (01:50 +0000)]

4 years agoAdded a comment: new backends
Ilya_Shlyakhter [Sun, 17 Apr 2022 19:02:10 +0000 (19:02 +0000)]
Added a comment: new backends

4 years agoDescribe compatibility bug with uftp 5.0
david@62eef13b7e480dc51c94ea50b67ebc0181410cd1 [Sat, 16 Apr 2022 22:29:25 +0000 (22:29 +0000)]
Describe compatibility bug with uftp 5.0

4 years agoAdded a comment: The man page export seems to be broken
lh [Fri, 15 Apr 2022 12:26:20 +0000 (12:26 +0000)]
Added a comment: The man page export seems to be broken

4 years agoAdded a comment
lell [Fri, 15 Apr 2022 12:05:19 +0000 (12:05 +0000)]
Added a comment

4 years agoAdded a comment: Myself
Abdulbasi [Thu, 14 Apr 2022 00:56:59 +0000 (00:56 +0000)]
Added a comment: Myself

4 years agoAdded a comment: recording state in git
lh [Wed, 13 Apr 2022 23:56:09 +0000 (23:56 +0000)]
Added a comment: recording state in git

4 years agoAdded a comment: Discrepancy between `git add` and `git annex add` when annex.largefi...
lh [Wed, 13 Apr 2022 23:15:25 +0000 (23:15 +0000)]
Added a comment: Discrepancy between `git add` and `git annex add` when annex.largefiles is set

4 years agoAdded a comment: Improving on the WORM situation
lh [Wed, 13 Apr 2022 22:19:40 +0000 (22:19 +0000)]
Added a comment: Improving on the WORM situation

4 years agotodo
Joey Hess [Wed, 13 Apr 2022 15:08:33 +0000 (11:08 -0400)]
todo

4 years agotodo
Joey Hess [Wed, 13 Apr 2022 03:26:17 +0000 (23:26 -0400)]
todo

4 years agoAdded a comment: maxextensionlength
Ilya_Shlyakhter [Fri, 8 Apr 2022 19:13:38 +0000 (19:13 +0000)]
Added a comment: maxextensionlength

4 years agoadd easy-git-annex to related_software
Joey Hess [Wed, 6 Apr 2022 17:13:35 +0000 (13:13 -0400)]
add easy-git-annex to related_software

4 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Wed, 6 Apr 2022 17:13:08 +0000 (13:13 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

4 years ago(no commit message)
jstritch [Wed, 6 Apr 2022 14:03:00 +0000 (14:03 +0000)]

4 years agohow to implement this?
Joey Hess [Tue, 5 Apr 2022 18:52:09 +0000 (14:52 -0400)]
how to implement this?

4 years agoadd a link to git-annex-dead
Joey Hess [Tue, 5 Apr 2022 18:47:27 +0000 (14:47 -0400)]
add a link to git-annex-dead

4 years agocomment
Joey Hess [Tue, 5 Apr 2022 17:44:07 +0000 (13:44 -0400)]
comment

4 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Tue, 5 Apr 2022 17:42:41 +0000 (13:42 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

4 years agoupdate
Joey Hess [Tue, 5 Apr 2022 13:35:47 +0000 (09:35 -0400)]
update

4 years agoAdded a comment
yarikoptic [Mon, 4 Apr 2022 20:01:14 +0000 (20:01 +0000)]
Added a comment

4 years agotodo triage
Joey Hess [Mon, 4 Apr 2022 19:22:49 +0000 (15:22 -0400)]
todo triage

Tagging todos that seem to have a plan ready as confirmed.

Also closed some old ones for various reasons. Including several that
turn out to be addressed by newer features.

Also opened a new todo about git-annex-config needing a criteria to add
new configs to it.

4 years agocomment
Joey Hess [Mon, 4 Apr 2022 18:27:48 +0000 (14:27 -0400)]
comment

4 years agocomment
Joey Hess [Mon, 4 Apr 2022 16:18:05 +0000 (12:18 -0400)]
comment

4 years agocommnet
Joey Hess [Mon, 4 Apr 2022 16:05:21 +0000 (12:05 -0400)]
commnet

4 years agoAdded a comment: TL;DR
amerlyq [Sat, 2 Apr 2022 14:13:48 +0000 (14:13 +0000)]
Added a comment: TL;DR

4 years agoAdded a comment: Inode rotation on android /sdcard
amerlyq [Sat, 2 Apr 2022 14:00:45 +0000 (14:00 +0000)]
Added a comment: Inode rotation on android /sdcard

4 years ago(no commit message)
Atemu [Fri, 1 Apr 2022 10:30:48 +0000 (10:30 +0000)]

4 years agoAdded a comment
Atemu [Fri, 1 Apr 2022 07:53:16 +0000 (07:53 +0000)]
Added a comment

4 years agoassistant prompt pushing of manual commits to remotes
Joey Hess [Thu, 31 Mar 2022 17:02:16 +0000 (13:02 -0400)]
assistant prompt pushing of manual commits to remotes

assistant: When annex.autocommit is set, notice commits that the user makes
manually, and push them out to remotes promptly.

Sponsored-by: Boyd Stephen Smith Jr. on Patreon
4 years agoAdded a comment
Atemu [Thu, 31 Mar 2022 10:39:28 +0000 (10:39 +0000)]
Added a comment

4 years agocomment
Joey Hess [Wed, 30 Mar 2022 16:56:57 +0000 (12:56 -0400)]
comment

4 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Wed, 30 Mar 2022 16:54:25 +0000 (12:54 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

4 years agomore thoughts
Joey Hess [Wed, 30 Mar 2022 16:53:35 +0000 (12:53 -0400)]
more thoughts

This idea seems fleshed out enough to implement now.

Sponsored-by: Boyd Stephen Smith Jr. on Patreon
4 years agoidea
Joey Hess [Wed, 30 Mar 2022 16:28:25 +0000 (12:28 -0400)]
idea

4 years agoAdded a comment
Atemu [Wed, 30 Mar 2022 13:10:48 +0000 (13:10 +0000)]
Added a comment

4 years agoAdded a comment: No bueno
https://openid.stackexchange.com/user/2098ac7f-d1c7-413d-8a8a-b7d0d7704ea3 [Tue, 29 Mar 2022 22:29:21 +0000 (22:29 +0000)]
Added a comment: No bueno

4 years agoidea
Joey Hess [Tue, 29 Mar 2022 22:09:41 +0000 (18:09 -0400)]
idea

4 years agoimprove documentation of overridding git-annex config with gitconfig
Joey Hess [Tue, 29 Mar 2022 19:03:57 +0000 (15:03 -0400)]
improve documentation of overridding git-annex config with gitconfig

4 years agocomment
Joey Hess [Tue, 29 Mar 2022 18:53:07 +0000 (14:53 -0400)]
comment

4 years agoretitle
Joey Hess [Tue, 29 Mar 2022 18:03:11 +0000 (14:03 -0400)]
retitle

4 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Mon, 28 Mar 2022 19:23:00 +0000 (15:23 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

4 years agoprevent numcopies or mincopies being configured to 0
Joey Hess [Mon, 28 Mar 2022 19:19:52 +0000 (15:19 -0400)]
prevent numcopies or mincopies being configured to 0

Ignore annex.numcopies set to 0 in gitattributes or git config, or by
git-annex numcopies or by --numcopies, since that configuration would make
git-annex easily lose data. Same for mincopies.

This is a continuation of the work to make data only be able to be lost
when --force is used. It earlier led to the --trust option being disabled,
and similar reasoning applies here.

Most numcopies configs had docs that strongly discouraged setting it to 0
anyway. And I can't imagine a use case for setting to 0. Not that there
might not be one, but it's just so far from the intended use case of
git-annex, of managing and storing your data, that it does not seem like
it makes sense to cater to such a hypothetical use case, where any
git-annex drop can lose your data at any time.

Using a smart constructor makes sure every place avoids 0. Note that this
does mean that NumCopies is for the configured desired values, and not the
actual existing number of copies, which of course can be 0. The name
configuredNumCopies is used to make that clear.

Sponsored-by: Brock Spratlen on Patreon
4 years ago(no commit message)
dev@c1c358f0d3c8563701193b66791eb1bc57a25ac9 [Sun, 27 Mar 2022 14:02:53 +0000 (14:02 +0000)]

4 years ago(no commit message)
dev@c1c358f0d3c8563701193b66791eb1bc57a25ac9 [Sun, 27 Mar 2022 14:02:20 +0000 (14:02 +0000)]

4 years agoswitch from hothasktags to hasktags
Joey Hess [Thu, 24 Mar 2022 17:07:49 +0000 (13:07 -0400)]
switch from hothasktags to hasktags

hothasktags was removed from Debian. And in the meantime hasktags
improved so it works well in the source tree.

Also make dev update tags.

Sponsored-by: Jack Hill on Patreon
4 years agoadd news item for git-annex 10.20220322
Joey Hess [Tue, 22 Mar 2022 17:56:54 +0000 (13:56 -0400)]
add news item for git-annex 10.20220322

4 years agoreleasing package git-annex version 10.20220322
Joey Hess [Tue, 22 Mar 2022 17:56:45 +0000 (13:56 -0400)]
releasing package git-annex version 10.20220322

4 years agofix quickcheck test on windows
Joey Hess [Tue, 22 Mar 2022 17:53:55 +0000 (13:53 -0400)]
fix quickcheck test on windows

prop_relPathDirToFileAbs_basics (TestableFilePath ":/") failed on
windows. The colon was filtered out after trying to make
the path relative, which only removed leading path separators.
So, ":/" changed to "/" which is not relative. Filtering out the colon
before hand avoids this problem.

Sponsored-by: Luke Shumaker on Patreon
4 years agoprop_relPathDirToFileAbs_basics has another hickup again in QuickChecks
jkniiv [Tue, 22 Mar 2022 17:15:15 +0000 (17:15 +0000)]
prop_relPathDirToFileAbs_basics has another hickup again in QuickChecks

4 years agoremove copyright entry for removed vendored library
Joey Hess [Tue, 22 Mar 2022 15:54:53 +0000 (11:54 -0400)]
remove copyright entry for removed vendored library

4 years agoRemoved the NetworkBSD build flag
Joey Hess [Tue, 22 Mar 2022 15:52:24 +0000 (11:52 -0400)]
Removed the NetworkBSD build flag

Debian stable and the i386ancient build both have a new enough network
to not need this flag any longer.

Sponsored-by: Svenne Krap on Patreon
4 years agoremove vendored http-client-restricted
Joey Hess [Tue, 22 Mar 2022 15:50:06 +0000 (11:50 -0400)]
remove vendored http-client-restricted

Removed vendored copy of http-client-restricted, and removed the
HttpClientRestricted build flag that avoided that dependency.

http-client-restricted is in Debian stable, and the i386ancient build also
uses it, so I think this vendored copy is no longer needed.

Sponsored-by: Noam Kremen on Patreon
4 years agobump i386ancient deps to allow using http-client-restricted and git-lfs
Joey Hess [Tue, 22 Mar 2022 15:36:48 +0000 (11:36 -0400)]
bump i386ancient deps to allow using http-client-restricted and git-lfs

This should let i386ancient limp along for a few years more, beyond the
removal of those vendored deps from git-annex.

Also networkbsd is set now, so probably the last thing to unset that
flag is gone, and the flag could be removed soon.

Sponsored-by: Jarkko Kniivilä on Patreon
4 years agorelax enough i386ancient deps to allow new tasty
Joey Hess [Tue, 22 Mar 2022 14:59:22 +0000 (10:59 -0400)]
relax enough i386ancient deps to allow new tasty

The new ansi-terminal was needed for test concurrency, and the new
concurrent-output fixes several bugs. And it turns out this is all
that's needed to use the new tasty.

Sponsored-by: Kevin Mueller on Patreon
4 years agore-relax tasty dependency version for i386ancient build
Joey Hess [Tue, 22 Mar 2022 14:31:10 +0000 (10:31 -0400)]
re-relax tasty dependency version for i386ancient build

Dependency issues were looking difficult to support tasty-1.2 with that
build. Not using `after` only affects rerunning and limiting tests,
since tasty's concurrency is not used, so this build will just not
support that.

We are probably nearing end of life on this build; it also doesn't
support git-lfs or http-client-restricted. The 2.6.32 kernel it supports
is at this point 13 years old, and stopped being supported by linux LTS
developers 10 years ago. It was supported by RHEL 6.10 through November
2020. At this point, no new hardware should be shipping with this
kernel, but that probably does not stop certian embedded vendors from
shipping it. And there is certainly some hardware still using it. But
the returns from supporting it are diminishing, and the quality of the
build for it is also diminishing.

Sponsored-by: Nicholas Golder-Manning on Patreon
4 years agobump ansi-terminal for tastyversion bump
Joey Hess [Tue, 22 Mar 2022 14:18:17 +0000 (10:18 -0400)]
bump ansi-terminal for tastyversion bump

4 years agoupdate tasty version to new minimum
Joey Hess [Mon, 21 Mar 2022 20:08:36 +0000 (16:08 -0400)]
update tasty version to new minimum

4 years agoreorder
Joey Hess [Mon, 21 Mar 2022 20:02:24 +0000 (16:02 -0400)]
reorder

4 years agocomment
Joey Hess [Mon, 21 Mar 2022 20:00:24 +0000 (16:00 -0400)]
comment

4 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Mon, 21 Mar 2022 19:58:50 +0000 (15:58 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

4 years agofix reversion
Joey Hess [Mon, 21 Mar 2022 19:50:27 +0000 (15:50 -0400)]
fix reversion

add: Avoid unncessarily converting a newly unlocked file to be stored
in git when it is not modified, even when annex.largefiles does not
match it.

This fixes a reversion in version 10.20220222, where git-annex unlock
followed by git-annex add, followed by git commit file could result in
git thinking the file was modified after the commit.

I do have half a mind to remove the withUnmodifiedUnlockedPointers part
of git-annex add. It seems weird, despite that old bug report arguing
a case of consistency that it ought to behave that way. When git-annex
add surpises me, it seems likely it's wrong.. But for now, this is the
smallest possible fix.

Sponsored-by: Dartmouth College's Datalad project
4 years agoAdded a comment
bz4 [Mon, 21 Mar 2022 19:18:53 +0000 (19:18 +0000)]
Added a comment

4 years agoAdded a comment
michael@ff03af62c7fd492c75066bda2fbf02370f5431f4 [Mon, 21 Mar 2022 19:10:35 +0000 (19:10 +0000)]
Added a comment

4 years agocomment
Joey Hess [Mon, 21 Mar 2022 17:29:56 +0000 (13:29 -0400)]
comment

4 years agodiagnose user error
Joey Hess [Mon, 21 Mar 2022 17:24:04 +0000 (13:24 -0400)]
diagnose user error

4 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Mon, 21 Mar 2022 17:12:40 +0000 (13:12 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

4 years agoadd back inode to directory special remote ContentIdentifier
Joey Hess [Mon, 21 Mar 2022 17:12:02 +0000 (13:12 -0400)]
add back inode to directory special remote ContentIdentifier

Directory special remotes with importtree=yes have changed to once more
take inodes into account. This will cause extra work when importing from a
directory on a FAT filesystem that changes inodes on every mount.

To avoid that extra work, set ignoreinodes=yes when initializing a new
directory special remote, or change the configuration of your existing
remote: git-annex enableremote foo ignoreinodes=yes

This will mean a one-time re-import of all contents from every directory
special remote due to the changed setting.

73df633a6215faa093e4a7524c6d328aa988aed1 thought
it was too unlikely that there would be modifications that the inode number
was needed to notice. That was probably right; it's very unlikely that a
file will get modified and end up with the same size and mtime as before.
But, what was not considered is that a program like NextCloud might write
two files with different content so closely together that they share the
mtime. The inode is necessary to detect that situation.

Sponsored-by: Max Thoursie on Patreon
4 years agoremoved
Lukey [Mon, 21 Mar 2022 17:00:33 +0000 (17:00 +0000)]
removed

4 years ago(no commit message)
bz4 [Mon, 21 Mar 2022 12:43:08 +0000 (12:43 +0000)]

4 years agoAdded a comment: Site Migration Issue Faced
Jamie [Mon, 21 Mar 2022 10:34:07 +0000 (10:34 +0000)]
Added a comment: Site Migration Issue Faced

4 years agoCreated
michael@ff03af62c7fd492c75066bda2fbf02370f5431f4 [Sun, 20 Mar 2022 15:30:31 +0000 (15:30 +0000)]
Created

4 years agoAdded a comment
beryllium@5bc3c32eb8156390f96e363e4ba38976567425ec [Fri, 18 Mar 2022 08:37:23 +0000 (08:37 +0000)]
Added a comment

4 years agoAdded a comment
rafd [Thu, 17 Mar 2022 16:09:38 +0000 (16:09 +0000)]
Added a comment

4 years ago(no commit message)
tag@e36c1c00f7653e4441059bbc5b3405e961379108 [Thu, 17 Mar 2022 14:13:58 +0000 (14:13 +0000)]

4 years ago(no commit message)
tag@e36c1c00f7653e4441059bbc5b3405e961379108 [Thu, 17 Mar 2022 14:13:16 +0000 (14:13 +0000)]

4 years agoAdded a comment
yarikoptic [Wed, 16 Mar 2022 20:32:57 +0000 (20:32 +0000)]
Added a comment

4 years agocomment
Joey Hess [Wed, 16 Mar 2022 19:18:24 +0000 (15:18 -0400)]
comment

4 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Wed, 16 Mar 2022 19:07:38 +0000 (15:07 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

4 years agotest: Added --jobs option
Joey Hess [Wed, 16 Mar 2022 18:42:07 +0000 (14:42 -0400)]
test: Added --jobs option

Default to the number of CPU cores, which seems about optimal
on my laptop. Using one more saves me 2 seconds actually.

Better packing of workers improves speed significantly.

In 2 tests runs, I saw segfaulting workers despite my attempt
to work around that issue. So detect when a worker does, and re-run it.

Removed installSignalHandlers again, because I was seeing an
error "lost signal due to full pipe", which I guess was somehow caused
by using it.

Sponsored-by: Dartmouth College's Datalad project
4 years agocleaner test dependencies
Joey Hess [Wed, 16 Mar 2022 16:53:08 +0000 (12:53 -0400)]
cleaner test dependencies

This improves the display of tests.

tasty-1.2 is in debian stable.

Sponsored-by: Dartmouth College's Datalad project
4 years agoclean up concurrent output of tests
Joey Hess [Wed, 16 Mar 2022 16:37:09 +0000 (12:37 -0400)]
clean up concurrent output of tests

Using concurrent-output this is easy. Just have to check if tasty has
color enabled, and propagate it into the worker processes, some of which
will be run without a controlling console.

Also added a call to installSignalHandlers; I noticed that interrupting
the test suite could leave the console in a bad state and this fixes
that.

The ansi-terminal dependency is free, since tasty also depends on it.

Sponsored-by: Dartmouth College's Datalad project
4 years agoAdded a comment: I have seen this
Gus [Wed, 16 Mar 2022 16:10:52 +0000 (16:10 +0000)]
Added a comment: I have seen this

4 years agoAdded a comment: FWIW a detail
yarikoptic [Wed, 16 Mar 2022 14:38:45 +0000 (14:38 +0000)]
Added a comment: FWIW a detail

4 years ago(no commit message)
bpoldrack [Wed, 16 Mar 2022 10:22:34 +0000 (10:22 +0000)]

4 years ago(no commit message)
bpoldrack [Wed, 16 Mar 2022 10:04:39 +0000 (10:04 +0000)]

4 years agolink to issue for tasty segfault
Joey Hess [Tue, 15 Mar 2022 23:30:57 +0000 (19:30 -0400)]
link to issue for tasty segfault

4 years ago(no commit message)
rafd [Tue, 15 Mar 2022 16:57:04 +0000 (16:57 +0000)]

4 years agoallow tests to be split for more granularity
Joey Hess [Mon, 14 Mar 2022 21:23:52 +0000 (17:23 -0400)]
allow tests to be split for more granularity

Unit tests are the main bulk of runtime, so splitting them into 2 or 3
parts should help.

For now, the number of parts is still 1, because on my 4 core laptop,
2 was a little bit slower, and 3 slower yet. However, this probably does
vary based on the number of cores, so needs to be revisited, and perhaps
made dynamic.

Since each test mode gets split into the specified number of parts,
plus property and remote tests, 2 gives 8 parts, and 3 gives 11 parts.
Load went to maybe 18, so there was probably contention slowing things
down.

So probably it needs to start N workers with some parts, and when a
worker finishes, run it with the next part, until all parts are
processed.

Sponsored-by: Dartmouth College's Datalad project
4 years agoparallelize git-annex test for 25% speedup
Joey Hess [Mon, 14 Mar 2022 19:24:37 +0000 (15:24 -0400)]
parallelize git-annex test for 25% speedup

Note the very weird workaround for what appears to be some kind of tasty
bug, which causes a segfault. This is not new to this modification,
I was seeing a segfault before at least intermittently when limiting
git-annex test -p to only run a single test group.

Also, the path from one test repo to a remote test repo used to be
"../../foo", which somehow broke when moving the test repos from .t to
.t/N. I don't actually quite understand how it used to work, but
"../foo" seems correct and works in the new situation.

Test output from the concurrent processes is not yet serialized.
Should be easy to do using concurrent-output.

More test groups will probably make the speedup larger. It would
probably be best to have a larger number of test groups and divvy them
amoung subprocesses numbered based on the number of CPU cores, perhaps
times 2 or 3.

Sponsored-by: Dartmouth College's Datalad project
4 years agofix accidental change to Test.hs while making comment
Joey Hess [Mon, 14 Mar 2022 17:10:37 +0000 (13:10 -0400)]
fix accidental change to Test.hs while making comment

4 years agoindentation
Joey Hess [Mon, 14 Mar 2022 16:30:27 +0000 (12:30 -0400)]
indentation

4 years agoAdded a comment
git-annex.visiteur@e9d364191d2ffc1b163c8d9e4c57dbadf58aad8e [Fri, 11 Mar 2022 18:46:40 +0000 (18:46 +0000)]
Added a comment