git-annex.git
16 months agoAdded a comment
Doable8234 [Mon, 16 Dec 2024 08:24:32 +0000 (08:24 +0000)]
Added a comment

16 months agoAdded a comment
Doable8234 [Mon, 16 Dec 2024 08:20:44 +0000 (08:20 +0000)]
Added a comment

16 months agoAdded a comment
matrss [Sun, 15 Dec 2024 23:39:34 +0000 (23:39 +0000)]
Added a comment

16 months agoAdded a comment
matrss [Sun, 15 Dec 2024 18:13:00 +0000 (18:13 +0000)]
Added a comment

16 months ago(no commit message)
eugen [Sat, 14 Dec 2024 17:48:08 +0000 (17:48 +0000)]

16 months agoAdded a comment
Doable8234 [Sat, 14 Dec 2024 08:15:22 +0000 (08:15 +0000)]
Added a comment

16 months agoAdded a comment
matrss [Fri, 13 Dec 2024 22:02:15 +0000 (22:02 +0000)]
Added a comment

16 months agodocument empty expression
Joey Hess [Fri, 13 Dec 2024 17:31:21 +0000 (13:31 -0400)]
document empty expression

16 months agoempty preferred content
Joey Hess [Fri, 13 Dec 2024 17:26:48 +0000 (13:26 -0400)]
empty preferred content

* Document that settting preferred content to "" is the same as the
  default unset behavior.
* sync: Avoid misleading warning about future preferred content
  transition when preferred content is set to "".

16 months agoAdded a comment
Doable8234 [Fri, 13 Dec 2024 02:04:40 +0000 (02:04 +0000)]
Added a comment

16 months agoresponse
Joey Hess [Thu, 12 Dec 2024 18:29:29 +0000 (14:29 -0400)]
response

16 months agocomment
Joey Hess [Thu, 12 Dec 2024 18:24:57 +0000 (14:24 -0400)]
comment

16 months ago(no commit message)
Doable8234 [Thu, 12 Dec 2024 05:16:49 +0000 (05:16 +0000)]

16 months ago(no commit message)
Doable8234 [Thu, 12 Dec 2024 04:57:37 +0000 (04:57 +0000)]

16 months ago(no commit message)
Doable8234 [Thu, 12 Dec 2024 04:53:22 +0000 (04:53 +0000)]

16 months agoAdded a comment: we use something like that..
vince@56a3a35e623d01b9236e911a9caff71eb367399b [Sun, 8 Dec 2024 20:34:35 +0000 (20:34 +0000)]
Added a comment: we use something like that..

16 months ago(no commit message)
vince@56a3a35e623d01b9236e911a9caff71eb367399b [Sun, 8 Dec 2024 20:26:18 +0000 (20:26 +0000)]

16 months ago(no commit message)
matrss [Sun, 8 Dec 2024 17:28:48 +0000 (17:28 +0000)]

16 months agoAdded a comment
matrss [Sun, 8 Dec 2024 17:27:59 +0000 (17:27 +0000)]
Added a comment

16 months ago(no commit message)
matrss [Sun, 8 Dec 2024 17:20:50 +0000 (17:20 +0000)]

16 months agoclose
Joey Hess [Wed, 4 Dec 2024 17:44:48 +0000 (13:44 -0400)]
close

16 months agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Wed, 4 Dec 2024 17:44:19 +0000 (13:44 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

16 months agocomments
Joey Hess [Tue, 3 Dec 2024 20:33:03 +0000 (16:33 -0400)]
comments

16 months agoAdded a comment
yarikoptic [Tue, 3 Dec 2024 20:26:10 +0000 (20:26 +0000)]
Added a comment

16 months agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Tue, 3 Dec 2024 18:39:30 +0000 (14:39 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

16 months agoannexInsteadOf config
Joey Hess [Tue, 3 Dec 2024 18:01:35 +0000 (14:01 -0400)]
annexInsteadOf config

Added config `url.<base>.annexInsteadOf` corresponding to git's
`url.<base>.pushInsteadOf`, to configure the urls to use for accessing the
git-annex repositories on a server without needing to configure
remote.name.annexUrl in each repository.

While one use case for this would be rewriting urls to use annex+http,
I decided not to add any kind of special case for that. So while
git-annex p2phttp, when serving multiple repositories, needs an url
of eg "annex+http://example.com/git-annex/ for each of them, rewriting an
url like "https://example.com/git/foo/bar" with this config set to
"https://example.com/git/" will result in eg
"annex+http://example.com/git-annex/foo/bar", which p2phttp does not
support.

That seems better dealt with in either git-annex p2phttp or a http
middleware, rather than complicating the config with a special case for
annex+http.

Anyway, there are other use cases for this that don't involve annex+http.

16 months agocomments
Joey Hess [Tue, 3 Dec 2024 17:00:14 +0000 (13:00 -0400)]
comments

16 months agoAdded a comment
iirekm [Tue, 3 Dec 2024 15:43:25 +0000 (15:43 +0000)]
Added a comment

16 months agocomment
Joey Hess [Tue, 3 Dec 2024 15:34:49 +0000 (11:34 -0400)]
comment

16 months ago(no commit message)
iirekm [Tue, 3 Dec 2024 15:28:22 +0000 (15:28 +0000)]

16 months agoAdded a comment: Maybe explanation
david@1439a1cab13195a56248b6a8fd98a62028bcba8a [Mon, 2 Dec 2024 20:25:34 +0000 (20:25 +0000)]
Added a comment: Maybe explanation

16 months agocomment
Joey Hess [Mon, 2 Dec 2024 18:07:43 +0000 (14:07 -0400)]
comment

16 months agocomment typo
Joey Hess [Mon, 2 Dec 2024 18:05:15 +0000 (14:05 -0400)]
comment typo

16 months agodeal with git's CFLR nonsense once again
Joey Hess [Mon, 2 Dec 2024 17:47:51 +0000 (13:47 -0400)]
deal with git's CFLR nonsense once again

Work around git hash-object --stdin-paths's odd stripping of carriage
return from the end of the line (some windows infection), avoiding crashing
when the repo contains a filename ending in a carriage return.

16 months agoadd news item for git-annex 10.20241202
Joey Hess [Mon, 2 Dec 2024 16:36:44 +0000 (12:36 -0400)]
add news item for git-annex 10.20241202

16 months agoreleasing package git-annex version 10.20241202
Joey Hess [Mon, 2 Dec 2024 16:36:24 +0000 (12:36 -0400)]
releasing package git-annex version 10.20241202

16 months agoforgot to add this comment earlier
Joey Hess [Mon, 2 Dec 2024 16:34:08 +0000 (12:34 -0400)]
forgot to add this comment earlier

16 months ago(no commit message)
zoran.bosnjak@683708e9d46ca9d9fa8957bd513e6648cbcbf421 [Sun, 1 Dec 2024 10:37:17 +0000 (10:37 +0000)]

16 months agoAdded a comment: thanks and the followup
psxvoid [Wed, 27 Nov 2024 09:08:32 +0000 (09:08 +0000)]
Added a comment: thanks and the followup

16 months ago(no commit message)
ljc [Wed, 27 Nov 2024 07:31:13 +0000 (07:31 +0000)]

16 months agocomment
Joey Hess [Mon, 25 Nov 2024 16:32:09 +0000 (12:32 -0400)]
comment

16 months agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Mon, 25 Nov 2024 16:16:32 +0000 (12:16 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

16 months agogit-remote-annex: Fix buggy behavior when annex.stalldetection is configured
Joey Hess [Mon, 25 Nov 2024 16:14:18 +0000 (12:14 -0400)]
git-remote-annex: Fix buggy behavior when annex.stalldetection is configured

Make programPath never return "git-remote-annex" or other known multi-call
program names, which are not git-annex and won't behave like it.
If the git-annex binary gets installed under some entirely other name,
it will still return it.

This change exposed that readProgramFile actually could crash,
which happened before only if getExecutablePath was not absolute
and there was no ~/.config/git-annex/program. So fixed that to catch
exception.

16 months agocomment
Joey Hess [Mon, 25 Nov 2024 15:33:38 +0000 (11:33 -0400)]
comment

16 months agoanalysis
Joey Hess [Mon, 25 Nov 2024 15:31:20 +0000 (11:31 -0400)]
analysis

16 months agoAdded a comment: Overriding git folder
aaron [Mon, 25 Nov 2024 02:55:25 +0000 (02:55 +0000)]
Added a comment: Overriding git folder

16 months agoAdded a comment: re: How to get a list of all NOT unused files
kyle [Mon, 25 Nov 2024 02:23:25 +0000 (02:23 +0000)]
Added a comment: re: How to get a list of all NOT unused files

16 months ago(no commit message)
aaron [Mon, 25 Nov 2024 01:07:20 +0000 (01:07 +0000)]

16 months ago(no commit message)
aaron [Mon, 25 Nov 2024 01:06:54 +0000 (01:06 +0000)]

16 months ago(no commit message)
aaron [Mon, 25 Nov 2024 01:05:32 +0000 (01:05 +0000)]

16 months ago(no commit message)
aaron [Mon, 25 Nov 2024 01:02:53 +0000 (01:02 +0000)]

16 months ago(no commit message)
aaron [Mon, 25 Nov 2024 00:56:41 +0000 (00:56 +0000)]

16 months ago(no commit message)
aaron [Mon, 25 Nov 2024 00:56:17 +0000 (00:56 +0000)]

16 months ago(no commit message)
aaron [Mon, 25 Nov 2024 00:55:07 +0000 (00:55 +0000)]

16 months agoshow remote name when failing
Joey Hess [Thu, 21 Nov 2024 19:55:32 +0000 (15:55 -0400)]
show remote name when failing

to help debug strange git behavior on some daily builds

16 months agomore fixing for building without servant
Joey Hess [Thu, 21 Nov 2024 19:35:06 +0000 (15:35 -0400)]
more fixing for building without servant

16 months agomore fixing for building without servant
Joey Hess [Thu, 21 Nov 2024 19:34:07 +0000 (15:34 -0400)]
more fixing for building without servant

16 months agocomment
Joey Hess [Thu, 21 Nov 2024 19:21:08 +0000 (15:21 -0400)]
comment

16 months agoMerge branch 'p2phttp-multi'
Joey Hess [Thu, 21 Nov 2024 19:16:06 +0000 (15:16 -0400)]
Merge branch 'p2phttp-multi'

16 months agop2phttp: notice when new repositories are added to --directory
Joey Hess [Thu, 21 Nov 2024 19:09:12 +0000 (15:09 -0400)]
p2phttp: notice when new repositories are added to --directory

When a uuid is not known, rescan for new repositories. Easy.

When a repository is removed, it will also get removed from the server
state on the next scan. But until a new uuid is seen, there will not be
a scan. This leaves the server trying to serve a uuid whose repository
is gone. That seems buggy. While getting just fails, dropping fails the
first time, but seems to leave the server in an unusable state, so the
next drop attempt hangs. The server is still able to serve other uuids,
only the one whose repository was removed has that problem.

16 months agoskip over repositories in --directory that do not have annex.uuid set
Joey Hess [Thu, 21 Nov 2024 18:18:18 +0000 (14:18 -0400)]
skip over repositories in --directory that do not have annex.uuid set

16 months agop2phttp support --jobs with --directory
Joey Hess [Thu, 21 Nov 2024 18:15:14 +0000 (14:15 -0400)]
p2phttp support --jobs with --directory

--jobs is usually an Annex option setter, but --directory runs in IO, so
would not have that available. So instead moved the option parser into
the command's Options.

16 months agop2phttp --directory implementation
Joey Hess [Thu, 21 Nov 2024 17:53:23 +0000 (13:53 -0400)]
p2phttp --directory implementation

Untested, but it compiles, so.

Known problems:

* --jobs is not available to startIO
* Does not notice when new repositories are added to a directory.
* Does not notice when repositories are removed from a directory.

16 months agomove the p2phttp server state map into a data type
Joey Hess [Thu, 21 Nov 2024 16:24:14 +0000 (12:24 -0400)]
move the p2phttp server state map into a data type

17 months agofix to build w/o servant
Joey Hess [Wed, 20 Nov 2024 20:29:43 +0000 (16:29 -0400)]
fix to build w/o servant

17 months agofix windows build
Joey Hess [Wed, 20 Nov 2024 19:19:23 +0000 (15:19 -0400)]
fix windows build

17 months agoin test suite display error from git push that fails to exit nonzero
Joey Hess [Wed, 20 Nov 2024 19:12:23 +0000 (15:12 -0400)]
in test suite display error from git push that fails to exit nonzero

17 months agogit-remote-annex: Require git version 2.31 or newer
Joey Hess [Wed, 20 Nov 2024 19:00:17 +0000 (15:00 -0400)]
git-remote-annex: Require git version 2.31 or newer

Since old ones had a buggy git bundle command.

In particular, git 2.30.2 has a git bundle that supports --stdin, but does
not read from it, and so fails to create a bundle.

While not using --stdin would perhaps work, it limits the number of revs
that get included in the bundle to the command line length limit.

But the real kicker is that at the same time --stdin got fixed, a bug also
got fixed that made git bundle skip including refs when they had the same
sha as other refs it included. Which would lead to data loss. So best to
avoid that buggy thing.

17 months agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Wed, 20 Nov 2024 17:41:40 +0000 (13:41 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

17 months agocomment
Joey Hess [Wed, 20 Nov 2024 17:41:34 +0000 (13:41 -0400)]
comment

17 months agodocument p2phttp --directory
Joey Hess [Wed, 20 Nov 2024 17:40:38 +0000 (13:40 -0400)]
document p2phttp --directory

The option is not implemented yet.

17 months agoadd proxied uuids to http server state map
Joey Hess [Wed, 20 Nov 2024 17:18:25 +0000 (13:18 -0400)]
add proxied uuids to http server state map

This fixes support for proxying after last commit broke it.

Note that withP2PConnections is called at server startup, and so only
proxies seen at that point will appear in the map and be used. It was
already the case that a proxy added after p2phttp was running would not
be served.

I think that is possibly a bug, but at least this commit doesn't
introduce the problem, though it might make it harder to fix it.

As bugs go, it's probably not a big deal, because after all,
git configs needs to be set in the local repository, followed by
git-annex updateproxy being run, to set up proxying. If someone is doing
that, they can restart their http server I suppose.

17 months agop2pHttpApp with a map of UUIDs to server states
Joey Hess [Wed, 20 Nov 2024 16:51:25 +0000 (12:51 -0400)]
p2pHttpApp with a map of UUIDs to server states

This is early groundwork for making p2phttp support serving multiple
repositories from a single daemon.

So far only 1 repository is served still. And this commit breaks support
for proxying!

17 months agoAdded a comment
yarikoptic [Wed, 20 Nov 2024 00:22:44 +0000 (00:22 +0000)]
Added a comment

17 months agoreuse http url password for p2phttp url when on same host
Joey Hess [Tue, 19 Nov 2024 19:27:26 +0000 (15:27 -0400)]
reuse http url password for p2phttp url when on same host

When remote.name.annexUrl is an annex+http(s) url, that uses the same
hostname as remote.name.url, which is itself a http(s) url, they are
assumed to share a username and password.

This avoids unnecessary duplicate password prompts.

17 months agoupdate
Joey Hess [Tue, 19 Nov 2024 18:42:50 +0000 (14:42 -0400)]
update

17 months agocomments
Joey Hess [Tue, 19 Nov 2024 18:26:47 +0000 (14:26 -0400)]
comments

17 months agotag INM7
Joey Hess [Tue, 19 Nov 2024 18:12:11 +0000 (14:12 -0400)]
tag INM7

17 months agocomment
Joey Hess [Tue, 19 Nov 2024 17:12:43 +0000 (13:12 -0400)]
comment

17 months agoskip git-remote-annex tests on windows
Joey Hess [Tue, 19 Nov 2024 17:01:12 +0000 (13:01 -0400)]
skip git-remote-annex tests on windows

The NullSoftInstaller does not install git-remote-annex. For that
matter, it does not install git-annex-shell either. I don't know quite
how it would make sense to do so, without hard links.
It could contain 3 copies of the same binary.

17 months agosplit git-remote-annex test
Joey Hess [Tue, 19 Nov 2024 16:54:23 +0000 (12:54 -0400)]
split git-remote-annex test

17 months agoretitle OSX bug
Joey Hess [Tue, 19 Nov 2024 16:46:01 +0000 (12:46 -0400)]
retitle OSX bug

17 months agogit-remote-annex: Fix cloning from a special remote on a crippled filesystem
Joey Hess [Tue, 19 Nov 2024 16:43:51 +0000 (12:43 -0400)]
git-remote-annex: Fix cloning from a special remote on a crippled filesystem

Not initializing and so deleting the bundles only causes a little more work
on the first git fetch.

17 months agoadd git-remote-annex as a dep of the test target
Joey Hess [Tue, 19 Nov 2024 16:13:13 +0000 (12:13 -0400)]
add git-remote-annex as a dep of the test target

17 months agoinitial report on failing tests
yarikoptic [Mon, 18 Nov 2024 13:55:55 +0000 (13:55 +0000)]
initial report on failing tests

17 months agoupdate
Joey Hess [Fri, 15 Nov 2024 20:36:43 +0000 (16:36 -0400)]
update

17 months agoclose
Joey Hess [Fri, 15 Nov 2024 19:34:03 +0000 (15:34 -0400)]
close

17 months agocomments
Joey Hess [Fri, 15 Nov 2024 19:31:49 +0000 (15:31 -0400)]
comments

17 months agoavoid storing same filename repeatedly in versioned import from S3
Joey Hess [Fri, 15 Nov 2024 19:11:56 +0000 (15:11 -0400)]
avoid storing same filename repeatedly in versioned import from S3

Logically, this should make it need a lot less memory when files have
been changed many times. In my tests, it didn't seem to change memory
use at all. Unsure why, it is working. It's possible the Response is not
getting garbage collected due to pinning. But as far as I can see, all
parts of it that are retained get copied in a way that won't keep the
whole thing pinned in memory.

17 months agouse 80% less memory when importing from a versioned S3 bucket
Joey Hess [Fri, 15 Nov 2024 18:19:02 +0000 (14:19 -0400)]
use 80% less memory when importing from a versioned S3 bucket

Same idea as commit eb714c107ba81d805458c770db8d4f22ae12a077, but even
better, because a lot of the response is DeleteMarker, that can be garbage
collected now.

17 months agouse 20% less memory when listing unversioned S3 bucket
Joey Hess [Fri, 15 Nov 2024 17:24:13 +0000 (13:24 -0400)]
use 20% less memory when listing unversioned S3 bucket

17 months ago(no commit message)
lell [Fri, 15 Nov 2024 09:27:28 +0000 (09:27 +0000)]

17 months ago(no commit message)
lell [Fri, 15 Nov 2024 09:26:56 +0000 (09:26 +0000)]

17 months ago(no commit message)
lell [Fri, 15 Nov 2024 09:25:32 +0000 (09:25 +0000)]

17 months agoAdded a comment
matrss [Fri, 15 Nov 2024 08:54:07 +0000 (08:54 +0000)]
Added a comment

17 months ago(no commit message)
lell [Fri, 15 Nov 2024 08:44:55 +0000 (08:44 +0000)]

17 months agofixed
Joey Hess [Thu, 14 Nov 2024 20:16:56 +0000 (16:16 -0400)]
fixed

17 months agoS3 use last Key when there is no Marker element
Joey Hess [Thu, 14 Nov 2024 20:09:27 +0000 (16:09 -0400)]
S3 use last Key when there is no Marker element

Fix infinite loop and memory blowup when importing from an unversioned S3
bucket that is large enough to need pagination.

I don't think there actually ever will be a Marker element, a delimiter is
not set.

Probably this code path was never tested with pagination! Also the aws
library's lack of any docs made it easy to mess up.

Versioned buckets seem to not have the same problem. The API docs for
ListObjectVersions say that NextKeyMarker will always be provided when
paginating.

17 months agocomments
Joey Hess [Thu, 14 Nov 2024 19:27:00 +0000 (15:27 -0400)]
comments

17 months agocomment
Joey Hess [Thu, 14 Nov 2024 18:20:05 +0000 (14:20 -0400)]
comment