git-annex.git
3 years agoMerge tag '10.20230407'
Sean Whitton [Wed, 14 Jun 2023 10:25:31 +0000 (11:25 +0100)]
Merge tag '10.20230407'

tagging package git-annex version 10.20230407

3 years agoreleasing package git-annex version 10.20230407
Joey Hess [Fri, 7 Apr 2023 17:37:03 +0000 (13:37 -0400)]
releasing package git-annex version 10.20230407

3 years agoexpand
Joey Hess [Fri, 7 Apr 2023 17:13:16 +0000 (13:13 -0400)]
expand

3 years agoresponse
Joey Hess [Fri, 7 Apr 2023 17:00:56 +0000 (13:00 -0400)]
response

3 years agoresponse
Joey Hess [Fri, 7 Apr 2023 16:52:43 +0000 (12:52 -0400)]
response

3 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Fri, 7 Apr 2023 16:42:09 +0000 (12:42 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

3 years ago(no commit message)
jwodder [Fri, 7 Apr 2023 15:19:04 +0000 (15:19 +0000)]

3 years agoAdded a comment: Status of the import/export protocol implementation
mih [Fri, 7 Apr 2023 09:17:22 +0000 (09:17 +0000)]
Added a comment: Status of the import/export protocol implementation

3 years agotodo
Joey Hess [Wed, 5 Apr 2023 23:37:21 +0000 (19:37 -0400)]
todo

3 years agorestore old registerurl location tracking behavior
Joey Hess [Wed, 5 Apr 2023 20:59:44 +0000 (16:59 -0400)]
restore old registerurl location tracking behavior

registerurl: When an url is claimed by a special remote other than the web,
update location tracking for that special remote.

registerurl's behavior was changed in commit
451171b7c1eaccfd0f39d4ec1d64c6964613f55a, apparently accidentially to not
update location tracking except for the web.

This makes registerurl followed by unregisterurl not be a no-op, when the
url happens to be claimed by a remote other than the web. It is a noop when
the url is unclaimed except by the web. I don't like the inconsistency,
and wish that registerurl and unregisterurl never updated location
tracking, which would be more in keeping with them being plumbing.

But there is the fact that it used to behave this way, and also it was
inconsistent that it updated location tracking for the web but not for
other remotes, unlike addurl. And there's an argument that the user might
not know what remote to expect to claim an url, so would be considerably in
the dark when using registerurl. (Although they have to know what content
gets downloaded, since they specify a key..)

Sponsored-By: the NIH-funded NICEMAN (ReproNim TR&D3) project
3 years agohm
Joey Hess [Wed, 5 Apr 2023 20:36:18 +0000 (16:36 -0400)]
hm

3 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Wed, 5 Apr 2023 19:56:31 +0000 (15:56 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

3 years agoregisterurl, unregisterurl: Added --remote option
Joey Hess [Wed, 5 Apr 2023 19:46:51 +0000 (15:46 -0400)]
registerurl, unregisterurl: Added --remote option

This serves two purposes. --remote=web bypasses other special remotes that
claim the url, same as addurl --raw. And, specifying some other remote
allows making sure that an url is claimed by the remote you expect,
which makes then using setpresentkey not be fragile.

Sponsored-By: the NIH-funded NICEMAN (ReproNim TR&D3) project
3 years agoAdded a comment
yarikoptic [Wed, 5 Apr 2023 19:36:41 +0000 (19:36 +0000)]
Added a comment

3 years agocomments
Joey Hess [Wed, 5 Apr 2023 19:00:39 +0000 (15:00 -0400)]
comments

3 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Wed, 5 Apr 2023 15:04:55 +0000 (11:04 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

3 years agoupdate
Joey Hess [Wed, 5 Apr 2023 15:04:35 +0000 (11:04 -0400)]
update

3 years agoAdded a comment
jkniiv [Wed, 5 Apr 2023 09:42:12 +0000 (09:42 +0000)]
Added a comment

3 years agoAdded a comment
yarikoptic [Wed, 5 Apr 2023 01:03:38 +0000 (01:03 +0000)]
Added a comment

3 years agoAdded a comment
yarikoptic [Wed, 5 Apr 2023 00:30:00 +0000 (00:30 +0000)]
Added a comment

3 years agoAdded a comment
yarikoptic [Tue, 4 Apr 2023 20:15:59 +0000 (20:15 +0000)]
Added a comment

3 years agoimprove comments
Joey Hess [Tue, 4 Apr 2023 19:21:50 +0000 (15:21 -0400)]
improve comments

3 years agofixed
Joey Hess [Tue, 4 Apr 2023 19:19:25 +0000 (15:19 -0400)]
fixed

3 years agoRevert "override rather than setting user.name and user.email"
Joey Hess [Tue, 4 Apr 2023 19:15:02 +0000 (15:15 -0400)]
Revert "override rather than setting user.name and user.email"

This reverts commit 66eb63dd820ade940b26b8eb00759d1b0f9242a0.

git-annex init is the only thing that uses ensureCommit. So overriding
there will make later commits to the git-annex branch or by git-annex sync
fail.

It's ugly that git-annex init sets user.name and user.email, but it only
does it on systems that are badly configured.

3 years agoSupport user.useConfigOnly git config
Joey Hess [Tue, 4 Apr 2023 19:12:52 +0000 (15:12 -0400)]
Support user.useConfigOnly git config

When it's set and git cannot determine user.name or user.email, this will
result in git-annex init failing when committing to create the git-annex
branch. Other git-annex commands that commit can also fail.

Sponsored-by: Jack Hill on Patreon
3 years agooverride rather than setting user.name and user.email
Joey Hess [Tue, 4 Apr 2023 18:56:44 +0000 (14:56 -0400)]
override rather than setting user.name and user.email

Avoid setting user.name and user.email in the git config when git is unable
to detect them.

git-annex has good reason to want to ensure git commit succeeds when eg
committing to the git-annex branch. But it's not playing nice to set these
values where other commands can see them.

Sponsored-by: Brett Eisenberg on Patreon
3 years agocomment
Joey Hess [Tue, 4 Apr 2023 18:43:33 +0000 (14:43 -0400)]
comment

3 years agocomment
Joey Hess [Tue, 4 Apr 2023 18:31:25 +0000 (14:31 -0400)]
comment

3 years agoprobably fixed but I didn't actually reproduce it
Joey Hess [Tue, 4 Apr 2023 18:25:55 +0000 (14:25 -0400)]
probably fixed but I didn't actually reproduce it

3 years agoreadFileStrict to avoid laziness bug
Joey Hess [Tue, 4 Apr 2023 18:25:01 +0000 (14:25 -0400)]
readFileStrict to avoid laziness bug

Fix laziness bug introduced in last release that breaks use of
--unlock-present and --hide-missing adjusted branches.

Since there is a writeFile of the same file immediately after readFile, it
may still have the file open for read (or may have happened to read it
already and closed it).

I was not able to reproduce the problem in brief testing, but this seems
obvious.

Sponsored-by: Luke Shumaker on Patreona
3 years agocomment
Joey Hess [Tue, 4 Apr 2023 18:09:02 +0000 (14:09 -0400)]
comment

3 years agocomments
Joey Hess [Tue, 4 Apr 2023 17:52:23 +0000 (13:52 -0400)]
comments

3 years agocomment
Joey Hess [Tue, 4 Apr 2023 17:34:32 +0000 (13:34 -0400)]
comment

3 years agoAdded a comment: happens during sync too
jkniiv [Mon, 3 Apr 2023 11:48:10 +0000 (11:48 +0000)]
Added a comment: happens during sync too

3 years ago(no commit message)
gioele@678b7c03f524f2669b179b603f65352fcc16774e [Mon, 3 Apr 2023 09:05:45 +0000 (09:05 +0000)]

3 years agoCommit Debian 3.0 (quilt) metadata
Sean Whitton [Sat, 1 Apr 2023 17:58:04 +0000 (10:58 -0700)]
Commit Debian 3.0 (quilt) metadata

[dgit (10.7) quilt-fixup]

3 years agochangelog
Sean Whitton [Sat, 1 Apr 2023 17:57:56 +0000 (10:57 -0700)]
changelog

3 years agoexternal protocol VERSION 2
Joey Hess [Tue, 28 Mar 2023 21:00:08 +0000 (17:00 -0400)]
external protocol VERSION 2

Support VERSION 2 in the external special remote protocol, which is
identical to VERSION 1, but avoids external remote programs neededing to
work around the above bug. External remote program that support
exporttree=yes are recommended to be updated to send VERSION 2.

Sponsored-by: Kevin Mueller on Patreon
3 years agofix concurrency bug causing EXPORT to be sent to the wrong external
Joey Hess [Tue, 28 Mar 2023 19:21:10 +0000 (15:21 -0400)]
fix concurrency bug causing EXPORT to be sent to the wrong external

Fix bug that caused broken protocol to be used with external remotes that
use exporttree=yes. In some cases this could result in the wrong content
being exported to, or retrieved from the remote.

Sponsored-by: Nicholas Golder-Manning on Patreon
3 years agoAdded a comment
gioele@678b7c03f524f2669b179b603f65352fcc16774e [Sat, 1 Apr 2023 11:09:49 +0000 (11:09 +0000)]
Added a comment

3 years agoAdded a comment
gioele@678b7c03f524f2669b179b603f65352fcc16774e [Sat, 1 Apr 2023 11:06:26 +0000 (11:06 +0000)]
Added a comment

3 years agoAdded a comment
wolf480@8ad1ccdd08efc303a88f7e88c4e629be6637a44e [Sat, 1 Apr 2023 07:23:18 +0000 (07:23 +0000)]
Added a comment

3 years agoreporting an annoying registerurl issue not registering a URL
yarikoptic [Fri, 31 Mar 2023 22:36:02 +0000 (22:36 +0000)]
reporting an annoying registerurl issue not registering a URL

3 years agoinitial todo for adding --remote to registerurl
yarikoptic [Fri, 31 Mar 2023 22:15:26 +0000 (22:15 +0000)]
initial todo for adding --remote to registerurl

3 years agoAdded a comment
gioele@678b7c03f524f2669b179b603f65352fcc16774e [Fri, 31 Mar 2023 20:16:04 +0000 (20:16 +0000)]
Added a comment

3 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Fri, 31 Mar 2023 19:16:28 +0000 (15:16 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

3 years agoSped up sqlite inserts 2x when built with persistent 2.14.5.0
Joey Hess [Fri, 31 Mar 2023 18:34:18 +0000 (14:34 -0400)]
Sped up sqlite inserts 2x when built with persistent 2.14.5.0

https://github.com/yesodweb/persistent/issues/1457

Sponsored-by: Dartmouth College's DANDI project
3 years agoAdded a comment
zhongruoyu@80ae9772857666624009364c29f07c70beed46ac [Fri, 31 Mar 2023 18:32:29 +0000 (18:32 +0000)]
Added a comment

3 years agoclarify
Joey Hess [Fri, 31 Mar 2023 18:18:50 +0000 (14:18 -0400)]
clarify

3 years agofix link
Joey Hess [Fri, 31 Mar 2023 18:18:36 +0000 (14:18 -0400)]
fix link

3 years agoidea
Joey Hess [Fri, 31 Mar 2023 17:11:35 +0000 (13:11 -0400)]
idea

3 years agogit-annex.cabal: Prevent building with unix-compat 0.7
Joey Hess [Fri, 31 Mar 2023 16:52:23 +0000 (12:52 -0400)]
git-annex.cabal: Prevent building with unix-compat 0.7

Which removed System.PosixCompat.User.
See https://github.com/haskell-pkg-janitors/unix-compat/issues/3

Sponsored-by: Noam Kremen on Patreon
3 years agoAdded a comment
wolf480@8ad1ccdd08efc303a88f7e88c4e629be6637a44e [Thu, 30 Mar 2023 17:33:58 +0000 (17:33 +0000)]
Added a comment

3 years agoAdded a comment
wolf480@8ad1ccdd08efc303a88f7e88c4e629be6637a44e [Thu, 30 Mar 2023 17:33:10 +0000 (17:33 +0000)]
Added a comment

3 years ago(no commit message)
zhongruoyu@80ae9772857666624009364c29f07c70beed46ac [Wed, 29 Mar 2023 22:44:08 +0000 (22:44 +0000)]

3 years ago(no commit message)
ptilopteri [Wed, 29 Mar 2023 22:32:06 +0000 (22:32 +0000)]

3 years agoadd news item for git-annex 10.20230329
Joey Hess [Wed, 29 Mar 2023 20:09:15 +0000 (16:09 -0400)]
add news item for git-annex 10.20230329

3 years agoreleasing package git-annex version 10.20230329
Joey Hess [Wed, 29 Mar 2023 20:09:05 +0000 (16:09 -0400)]
releasing package git-annex version 10.20230329

3 years ago(no commit message)
sawmke [Wed, 29 Mar 2023 11:47:13 +0000 (11:47 +0000)]

3 years agoAdded a comment
dpifke [Wed, 29 Mar 2023 03:13:50 +0000 (03:13 +0000)]
Added a comment

3 years agofix whitespace
Joey Hess [Tue, 28 Mar 2023 21:13:04 +0000 (17:13 -0400)]
fix whitespace

3 years agoprep for release tomorrow
Joey Hess [Tue, 28 Mar 2023 21:02:34 +0000 (17:02 -0400)]
prep for release tomorrow

3 years agoclose
Joey Hess [Tue, 28 Mar 2023 21:01:25 +0000 (17:01 -0400)]
close

3 years agoexternal protocol VERSION 2
Joey Hess [Tue, 28 Mar 2023 21:00:08 +0000 (17:00 -0400)]
external protocol VERSION 2

Support VERSION 2 in the external special remote protocol, which is
identical to VERSION 1, but avoids external remote programs neededing to
work around the above bug. External remote program that support
exporttree=yes are recommended to be updated to send VERSION 2.

Sponsored-by: Kevin Mueller on Patreon
3 years agocomment
Joey Hess [Tue, 28 Mar 2023 20:08:46 +0000 (16:08 -0400)]
comment

3 years agofixed
Joey Hess [Tue, 28 Mar 2023 19:26:26 +0000 (15:26 -0400)]
fixed

3 years agofix concurrency bug causing EXPORT to be sent to the wrong external
Joey Hess [Tue, 28 Mar 2023 19:21:10 +0000 (15:21 -0400)]
fix concurrency bug causing EXPORT to be sent to the wrong external

Fix bug that caused broken protocol to be used with external remotes that
use exporttree=yes. In some cases this could result in the wrong content
being exported to, or retrieved from the remote.

Sponsored-by: Nicholas Golder-Manning on Patreon
3 years agoclarify EXPORT
Joey Hess [Tue, 28 Mar 2023 19:17:56 +0000 (15:17 -0400)]
clarify EXPORT

3 years agocomment
Joey Hess [Tue, 28 Mar 2023 18:31:00 +0000 (14:31 -0400)]
comment

3 years agocomment
Joey Hess [Tue, 28 Mar 2023 18:26:19 +0000 (14:26 -0400)]
comment

3 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Tue, 28 Mar 2023 18:09:22 +0000 (14:09 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

3 years agocomment
Joey Hess [Tue, 28 Mar 2023 18:08:24 +0000 (14:08 -0400)]
comment

3 years agofix comment
Joey Hess [Tue, 28 Mar 2023 17:40:17 +0000 (13:40 -0400)]
fix comment

3 years ago(no commit message)
jonahmail1@a6afaf85da848e3a3cf7fcec09326a982d4c1819 [Tue, 28 Mar 2023 17:27:46 +0000 (17:27 +0000)]

3 years ago(no commit message)
jonahmail1@a6afaf85da848e3a3cf7fcec09326a982d4c1819 [Tue, 28 Mar 2023 17:27:33 +0000 (17:27 +0000)]

3 years agocomment
Joey Hess [Tue, 28 Mar 2023 17:13:34 +0000 (13:13 -0400)]
comment

3 years agoCopy with a reflink when exporting a tree to a directory special remote
Joey Hess [Tue, 28 Mar 2023 17:06:11 +0000 (13:06 -0400)]
Copy with a reflink when exporting a tree to a directory special remote

Remote.Directory makes a temp file, then calls this, and since the temp
file exists, it prevented probing if CoW works.

Note that deleting the empty file does mean there's a small window for a
race. If another process is also exporting to the remote, that could let it
make the same temp file. However, the temp filename actually has the
processes's pid in it, which avoids that being a problem.

This may have been a reversion caused by commits around
63d508e8855b2e61a725c906ea17d1c7f4a2e125, but I haven't gone back and
tested to be sure. The directory special remote had supposedly supported
CoW for this going back to about half a year before that.

Sponsored-by: Graham Spencer on Patreon
3 years agofix comment
Joey Hess [Tue, 28 Mar 2023 16:38:47 +0000 (12:38 -0400)]
fix comment

3 years agocomment
Joey Hess [Tue, 28 Mar 2023 16:14:23 +0000 (12:14 -0400)]
comment

3 years agoAdded a comment: annex.bwlimit and jobs
gerzoyayde@85d55f6dec266753698f42d1c8e06917ef6674a3 [Tue, 28 Mar 2023 00:00:44 +0000 (00:00 +0000)]
Added a comment: annex.bwlimit and jobs

3 years agoaddurl, importfeed: Fix failure when annex.securehashesonly is set
Joey Hess [Mon, 27 Mar 2023 19:10:46 +0000 (15:10 -0400)]
addurl, importfeed: Fix failure when annex.securehashesonly is set

The temporary URL key used for the download, before the real key is
generated, was blocked by annex.securehashesonly.

Fixed by passing the Backend that will be used for the final key into
runTransfer. When a Backend is provided, have preCheckSecureHashes
check that, rather than the key being transferred.

Sponsored-by: unqueued on Patreon
3 years agoremove unused Key parameter from isCryptographicallySecure
Joey Hess [Mon, 27 Mar 2023 18:34:00 +0000 (14:34 -0400)]
remove unused Key parameter from isCryptographicallySecure

This will allow using isCryptographicallySecure on a Backend, before a
Key has been generated.

Sponsored-by: Lawrence Brogan on Patreon
3 years agopromote comment to bug
Joey Hess [Mon, 27 Mar 2023 18:10:32 +0000 (14:10 -0400)]
promote comment to bug

3 years agoverified fixed
Joey Hess [Mon, 27 Mar 2023 17:58:16 +0000 (13:58 -0400)]
verified fixed

3 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Mon, 27 Mar 2023 17:38:13 +0000 (13:38 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

3 years agoWindows: Support urls like "file:///c:/path"
Joey Hess [Mon, 27 Mar 2023 17:38:02 +0000 (13:38 -0400)]
Windows: Support urls like "file:///c:/path"

That is a legal url, but parseUrl parses it to "/c:/path"
which is not a valid path on Windows. So as a workaround, use
parseURIPortable everywhere, which removes the leading slash when
run on windows.

Note that if an url is parsed like this and then serialized back
to a string, it will be different from the input. Which could
potentially be a problem, but is probably not in practice.

An alternative way to do it would be to have an uriPathPortable
that fixes up the path after parsing. But it would be harder to
make sure that is used everywhere, since uriPath is also used
when constructing an URI.

It's also worth noting that System.FilePath.normalize "/c:/path"
yields "c:/path". The reason I didn't use it is that it also
may change "/" to "\" in the path and I wanted to keep the url
changes minimal. Also noticed that convertToWindowsNativeNamespace
handles "/c:/path" the same as "c:/path".

Sponsored-By: the NIH-funded NICEMAN (ReproNim TR&D3) project
3 years agocomment
Joey Hess [Mon, 27 Mar 2023 16:36:21 +0000 (12:36 -0400)]
comment

3 years agoavoid build warning on windows
Joey Hess [Mon, 27 Mar 2023 16:21:40 +0000 (12:21 -0400)]
avoid build warning on windows

3 years agoavoid build warning on windows
Joey Hess [Mon, 27 Mar 2023 16:20:35 +0000 (12:20 -0400)]
avoid build warning on windows

3 years agoavoid build warning on windows
Joey Hess [Mon, 27 Mar 2023 16:19:26 +0000 (12:19 -0400)]
avoid build warning on windows

3 years agofix build warning on windows
Joey Hess [Mon, 27 Mar 2023 16:17:55 +0000 (12:17 -0400)]
fix build warning on windows

3 years ago(no commit message)
jonas [Sun, 26 Mar 2023 20:04:22 +0000 (20:04 +0000)]

3 years ago(no commit message)
wolf480@8ad1ccdd08efc303a88f7e88c4e629be6637a44e [Sun, 26 Mar 2023 19:00:42 +0000 (19:00 +0000)]

3 years ago(no commit message)
wolf480@8ad1ccdd08efc303a88f7e88c4e629be6637a44e [Sun, 26 Mar 2023 18:39:20 +0000 (18:39 +0000)]

3 years agoAdded a comment
gioele@678b7c03f524f2669b179b603f65352fcc16774e [Sat, 25 Mar 2023 08:47:01 +0000 (08:47 +0000)]
Added a comment

3 years agoAdded a comment: securehashesonly conflicts with addurl
jt [Sat, 25 Mar 2023 03:22:50 +0000 (03:22 +0000)]
Added a comment: securehashesonly conflicts with addurl

3 years agoannex.maxextensionlength for view
Joey Hess [Fri, 24 Mar 2023 17:53:51 +0000 (13:53 -0400)]
annex.maxextensionlength for view

view: Support annex.maxextensionlength when generating filenames for the
view branch.

Note that refining an existing view will reuse the extension length that was
configured when initially constructing the view. This is necessarily the case
because it reuses the filenames.

Also view files used to have all extensions at the end, no matter how
many there were. Since annex.maxextensionlength's documentation includes
that it's limited to 2 extensions, I made it consistent with that.

Sponsored-by: k0ld on Patreon
3 years agoAdded a comment
gioele@678b7c03f524f2669b179b603f65352fcc16774e [Fri, 24 Mar 2023 08:13:00 +0000 (08:13 +0000)]
Added a comment

3 years agoAdded a comment
dpifke [Thu, 23 Mar 2023 23:43:17 +0000 (23:43 +0000)]
Added a comment

3 years agoAdded a comment
dpifke [Thu, 23 Mar 2023 23:32:45 +0000 (23:32 +0000)]
Added a comment