git-annex.git
5 years agoAdded a comment: Unless the key is the argument to set
seanl@fe5df935169a5440a52bdbfc5fece85cdd002d68 [Tue, 26 Jan 2021 16:12:31 +0000 (16:12 +0000)]
Added a comment: Unless the key is the argument to set

5 years agoAdded a comment: -- works between the value and the option
seanl@fe5df935169a5440a52bdbfc5fece85cdd002d68 [Tue, 26 Jan 2021 16:10:23 +0000 (16:10 +0000)]
Added a comment: -- works between the value and the option

5 years agoRemove broken link.
edward [Tue, 26 Jan 2021 16:01:58 +0000 (16:01 +0000)]
Remove broken link.

5 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Tue, 26 Jan 2021 15:32:52 +0000 (11:32 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

5 years agoupdate
Joey Hess [Tue, 26 Jan 2021 15:32:40 +0000 (11:32 -0400)]
update

5 years agoadd
Joey Hess [Tue, 26 Jan 2021 15:27:14 +0000 (11:27 -0400)]
add

5 years agoAdded a comment: Other benefits of proposed improvements to the smudge/clean API
hello@6d9a437cebceb2fc657f93c4f30a7ba859e9309c [Tue, 26 Jan 2021 14:25:59 +0000 (14:25 +0000)]
Added a comment: Other benefits of proposed improvements to the smudge/clean API

5 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Tue, 26 Jan 2021 01:51:28 +0000 (21:51 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

5 years agofix breakage caused by recent commit
Joey Hess [Tue, 26 Jan 2021 01:19:07 +0000 (21:19 -0400)]
fix breakage caused by recent commit

34a535ebeaf483f6643f949ae774939650d74837 broke the test suite.
Getting a file started failing in one case, because the annex object did
not have its inode cached, so was not trusted to be unmodified.

This adds something very similar to what was added to linkAnnex
in commit 2e9341a47d437b5a7b14824d3e5b2864b059a850 -- if there are not
yet any inodes cached for a key, add the inode of the annex object when
adding the inode of the unlocked file.

Feels like this should be handled in a more principled way. How
do we know the addInodeCaches call in getMoveRaceRecovery just above
this change is currently correct? It doesn't add the annex object inode
cache. Ah well, maybe sometime when I've not had my entire evening eaten
by a reversion that the test suite caught as I was cooking dinner.

5 years ago(no commit message)
seanl@fe5df935169a5440a52bdbfc5fece85cdd002d68 [Tue, 26 Jan 2021 01:05:19 +0000 (01:05 +0000)]

5 years ago(no commit message)
seanl@fe5df935169a5440a52bdbfc5fece85cdd002d68 [Tue, 26 Jan 2021 01:04:13 +0000 (01:04 +0000)]

5 years agowording
Joey Hess [Mon, 25 Jan 2021 21:40:08 +0000 (17:40 -0400)]
wording

5 years agosupport modifying and running git add on an unlocked file that used an URL key
Joey Hess [Mon, 25 Jan 2021 21:34:58 +0000 (17:34 -0400)]
support modifying and running git add on an unlocked file that used an URL key

Avoids the smudge --clean filter failing because URL keys do not support
genKey. Instead the modified content will be added using the default
backend.

This commit was sponsored by Jochen Bartl on Patreon.

5 years agoadjust: Fix some bad behavior when unlocked files use URL keys.
Joey Hess [Mon, 25 Jan 2021 21:25:42 +0000 (17:25 -0400)]
adjust: Fix some bad behavior when unlocked files use URL keys.

This avoids the smudge --clean filter failing on the URL keys.

git checkout runs the post-checkout hook, which runs smudge --update.
That populates all the pointer files, but it neglected to store their inode
caches in the keys db. With that done, and the keys db flushed before
smudge --clean gets run (by restagePointerFile), the isUnmodifiedCheap
check can tell the file is not modified, so will not try to re-ingest it,
which does not work with URL keys because they do not support genKey.

It also seems possible that the isUnmodifiedCheap was also failing for
non-URL keys, which would cause them to be re-ingested, leading to a lot of
extra work. I have not verified that, but don't see why it wouldn't have
happened. So this probably also speeds up checking out adjusted branches.

This commit was sponsored by Boyd Stephen Smith Jr. on Patreon.

5 years agobug
Joey Hess [Mon, 25 Jan 2021 20:08:30 +0000 (16:08 -0400)]
bug

5 years agofix export overwrite on FAT
Joey Hess [Mon, 25 Jan 2021 17:31:17 +0000 (13:31 -0400)]
fix export overwrite on FAT

Don't accept the cid of the temp file that the content has just been
written to as something we will accept if another file has that same
content. There's no reason to, and on FAT, due to mtime resolution,
the test suite hit just such a case.

This fixes a reversion from 73df633a6215faa093e4a7524c6d328aa988aed1
which removed inode from the ContentIdentifier.

5 years agoAdded a comment
yarikoptic [Mon, 25 Jan 2021 16:49:36 +0000 (16:49 +0000)]
Added a comment

5 years ago(no commit message)
jwodder [Mon, 25 Jan 2021 16:47:34 +0000 (16:47 +0000)]

5 years agocomment
Joey Hess [Mon, 25 Jan 2021 16:31:47 +0000 (12:31 -0400)]
comment

5 years agocomment
Joey Hess [Mon, 25 Jan 2021 16:04:07 +0000 (12:04 -0400)]
comment

5 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Mon, 25 Jan 2021 15:31:46 +0000 (11:31 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

5 years agonotabug
Joey Hess [Mon, 25 Jan 2021 15:20:46 +0000 (11:20 -0400)]
notabug

5 years agoAdded a comment
yarikoptic [Mon, 25 Jan 2021 15:18:40 +0000 (15:18 +0000)]
Added a comment

5 years agoAdded a comment
kyle [Mon, 25 Jan 2021 14:52:54 +0000 (14:52 +0000)]
Added a comment

5 years agoAdd positive stuff
seanl@fe5df935169a5440a52bdbfc5fece85cdd002d68 [Sun, 24 Jan 2021 23:24:32 +0000 (23:24 +0000)]
Add positive stuff

5 years ago(no commit message)
seanl@fe5df935169a5440a52bdbfc5fece85cdd002d68 [Sun, 24 Jan 2021 23:19:36 +0000 (23:19 +0000)]

5 years agoAdded a comment: bug report
m15 [Sat, 23 Jan 2021 08:52:14 +0000 (08:52 +0000)]
Added a comment: bug report

5 years agocomment
Joey Hess [Fri, 22 Jan 2021 18:56:18 +0000 (14:56 -0400)]
comment

5 years agofollowup
Joey Hess [Fri, 22 Jan 2021 18:36:34 +0000 (14:36 -0400)]
followup

5 years agofix problem on windows with newly rewritten prop_relPathDirToFileAbs_basics
Joey Hess [Fri, 22 Jan 2021 18:21:29 +0000 (14:21 -0400)]
fix problem on windows with newly rewritten prop_relPathDirToFileAbs_basics

Seems that dropDrive on windows only drops eg c:/ but not a leading /
while on linux, it does drop a leading / (which is what it considers
to be equivilant to a drive letter. I had been relying on it to drop
both. So need to drop leading directory separators.

Also, if the quickcheck generated input is eg "c:c:c:c:foo",
dropDrive will only drop the first one, leaving a path that's
still not relative. So instead of using dropDrive, just remove the
colons from the path.

5 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Fri, 22 Jan 2021 18:05:49 +0000 (14:05 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

5 years agoFix a bug in view filename generation when a metadata value ended with "/"
Joey Hess [Fri, 22 Jan 2021 18:05:14 +0000 (14:05 -0400)]
Fix a bug in view filename generation when a metadata value ended with "/"

Or ":" or "\" on Windows, eg "c:" again.

5 years agoupdate copyright year
Joey Hess [Fri, 22 Jan 2021 18:00:36 +0000 (14:00 -0400)]
update copyright year

5 years agoinitial report on export_import fail
yarikoptic [Fri, 22 Jan 2021 16:50:58 +0000 (16:50 +0000)]
initial report on export_import fail

5 years agoAdded a comment
yarikoptic [Fri, 22 Jan 2021 16:44:58 +0000 (16:44 +0000)]
Added a comment

5 years agoAdded a comment
yarikoptic [Fri, 22 Jan 2021 16:42:32 +0000 (16:42 +0000)]
Added a comment

5 years agoAdded a comment
mike@2d6d71f56ce2a992244350475251df87c26fe351 [Fri, 22 Jan 2021 11:08:38 +0000 (11:08 +0000)]
Added a comment

5 years agoAdded a comment
yarikoptic [Fri, 22 Jan 2021 01:47:46 +0000 (01:47 +0000)]
Added a comment

5 years agofix a bug that prevented git-annex init from working in a submodule
Joey Hess [Thu, 21 Jan 2021 19:33:15 +0000 (15:33 -0400)]
fix a bug that prevented git-annex init from working in a submodule

This is probably a reversion, but not sure what caused it. By the time
Annex.Init runs fixupUnusualReposAfterInit, another git-annex process has
at least sometimes already done the necessary fixups. (Eg, one run
indirectly by a git command.) But since the Repo is cached, it doesn't
realize and does them again. So, avoid crashing when git config --unset
fails.

This commit was sponsored by Jack Hill on Patreon.

5 years agocomment
Joey Hess [Thu, 21 Jan 2021 19:25:46 +0000 (15:25 -0400)]
comment

5 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Thu, 21 Jan 2021 19:19:46 +0000 (15:19 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

5 years agoAdded a comment
yarikoptic [Thu, 21 Jan 2021 18:36:50 +0000 (18:36 +0000)]
Added a comment

5 years agocomment typo
Joey Hess [Thu, 21 Jan 2021 18:13:55 +0000 (14:13 -0400)]
comment typo

5 years agochange user-visible error to giveup
Joey Hess [Thu, 21 Jan 2021 18:13:14 +0000 (14:13 -0400)]
change user-visible error to giveup

5 years agocomment
Joey Hess [Thu, 21 Jan 2021 17:19:40 +0000 (13:19 -0400)]
comment

5 years agoadjusted report with more of not so deep thoughts/observation ;)
yarikoptic [Thu, 21 Jan 2021 03:30:07 +0000 (03:30 +0000)]
adjusted report with more of not so deep thoughts/observation ;)

5 years agoinitial plea for considering credentials upon 404 of /config
yarikoptic [Thu, 21 Jan 2021 03:26:09 +0000 (03:26 +0000)]
initial plea for considering credentials upon 404 of /config

5 years ago(no commit message)
mark@6b90344cdab3158eacb94a3944460d138afc9bef [Wed, 20 Jan 2021 17:36:20 +0000 (17:36 +0000)]

5 years agoAdded a comment
mike@2d6d71f56ce2a992244350475251df87c26fe351 [Wed, 20 Jan 2021 15:41:50 +0000 (15:41 +0000)]
Added a comment

5 years agoAdded a comment
mike@2d6d71f56ce2a992244350475251df87c26fe351 [Tue, 19 Jan 2021 18:16:28 +0000 (18:16 +0000)]
Added a comment

5 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Tue, 19 Jan 2021 17:20:22 +0000 (13:20 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

5 years agoomit inode from ContentIdentifier for directory special remote
Joey Hess [Tue, 19 Jan 2021 16:57:15 +0000 (12:57 -0400)]
omit inode from ContentIdentifier for directory special remote

Directory special remotes with importtree=yes now avoid unncessary overhead
when inodes of files have changed, as happens whenever a FAT filesystem
gets remounted.

A few unusual edge cases of modifications won't be detected and
imported. I think they're unusual enough not to be a concern. It would
be possible to add a config setting that controls whether to compare
inodes too, but does not seem worth bothering the user about currently.

I chose to continue to use the InodeCache serialization, just with the
inode zeroed. This way, if I later change my mind or make it
configurable, can parse it back to an InodeCache and operate on it. The
overhead of storing a 0 in the content identifier log seems worth it.

There is a one-time cost to this change; all directory special remotes
with importtree=yes will re-hash all files once, and will update the
content identifier logs with zeroed inodes.

This commit was sponsored by Brett Eisenberg on Patreon.

5 years agoremoved
Lukey [Tue, 19 Jan 2021 16:49:57 +0000 (16:49 +0000)]
removed

5 years agoAdded a comment
Lukey [Tue, 19 Jan 2021 16:49:05 +0000 (16:49 +0000)]
Added a comment

5 years agoAdded a comment
Lukey [Tue, 19 Jan 2021 16:45:31 +0000 (16:45 +0000)]
Added a comment

5 years agodisplay exception as part of warnings
Joey Hess [Tue, 19 Jan 2021 16:27:42 +0000 (12:27 -0400)]
display exception as part of warnings

and comment that led to this change

5 years agoAdded a comment
Lukey [Tue, 19 Jan 2021 16:27:07 +0000 (16:27 +0000)]
Added a comment

5 years agocomment
Joey Hess [Tue, 19 Jan 2021 16:12:34 +0000 (12:12 -0400)]
comment

5 years agocomment
Joey Hess [Tue, 19 Jan 2021 16:11:38 +0000 (12:11 -0400)]
comment

5 years agocomment and todo
Joey Hess [Tue, 19 Jan 2021 15:56:14 +0000 (11:56 -0400)]
comment and todo

5 years agoremoved
mike@2d6d71f56ce2a992244350475251df87c26fe351 [Tue, 19 Jan 2021 12:21:38 +0000 (12:21 +0000)]
removed

5 years agoAdded a comment
mike@2d6d71f56ce2a992244350475251df87c26fe351 [Tue, 19 Jan 2021 12:21:07 +0000 (12:21 +0000)]
Added a comment

5 years agoAdded a comment
mike@2d6d71f56ce2a992244350475251df87c26fe351 [Tue, 19 Jan 2021 12:19:46 +0000 (12:19 +0000)]
Added a comment

5 years ago(no commit message)
mike@2d6d71f56ce2a992244350475251df87c26fe351 [Tue, 19 Jan 2021 12:04:48 +0000 (12:04 +0000)]

5 years agoAdded a comment: Went away with fsck?
seanl@fe5df935169a5440a52bdbfc5fece85cdd002d68 [Tue, 19 Jan 2021 01:15:51 +0000 (01:15 +0000)]
Added a comment: Went away with fsck?

5 years agoAdded a comment: Issue when using flag --recurse-submodules with git-checkout and...
aschoise@2386473665998adccbe71286036e57d7f3ad3a58 [Mon, 18 Jan 2021 23:04:17 +0000 (23:04 +0000)]
Added a comment: Issue when using flag --recurse-submodules with git-checkout and git-switch

5 years ago(no commit message)
rdp@af74e2cdae4a6bbe02b41cfc0821a1980ed59293 [Mon, 18 Jan 2021 22:31:29 +0000 (22:31 +0000)]

5 years agoavoid partial functions in Git.Url
Joey Hess [Mon, 18 Jan 2021 19:07:23 +0000 (15:07 -0400)]
avoid partial functions in Git.Url

After the last commit, it was able to throw errors just due to an
unparseable url. This avoids needing to worry about that, as long
as the call site has already checked that it has a parseable url.

5 years agoavoid crashing when there are remotes using unparseable urls
Joey Hess [Mon, 18 Jan 2021 18:52:56 +0000 (14:52 -0400)]
avoid crashing when there are remotes using unparseable urls

Including the non-standard URI form that git-remote-gcrypt uses for rsync.

Eg, "ook://foo:bar" cannot be parsed because "bar" is not a valid port
number. But git could have a remote with that, it would try to run
git-remote-ook to handle it. So, git-annex has to allow for such things,
rather than crashing.

This commit was sponsored by Luke Shumaker on Patreon.

5 years agocomment
Joey Hess [Mon, 18 Jan 2021 17:54:57 +0000 (13:54 -0400)]
comment

5 years agocomment
Joey Hess [Mon, 18 Jan 2021 17:40:31 +0000 (13:40 -0400)]
comment

5 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Mon, 18 Jan 2021 17:28:08 +0000 (13:28 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

5 years agofix prop_relPathDirToFileAbs_basics fail on windows
Joey Hess [Mon, 18 Jan 2021 17:19:17 +0000 (13:19 -0400)]
fix prop_relPathDirToFileAbs_basics fail on windows

It was just slapping on a path separator to the front of the path to
make it absolute, but on windows, a path like "//foo/bar" actually
has a network "drive" of "//foo" and so that broke the test case.

Since "a:foo" is a somehow relative path on windows
(who knows how), drop any drive from the input. But dropDrive also drops
any leading path separator, making the input path relative. So now
it should be safe to slapp on a leading path separator.

5 years agoAdded a comment
Lukey [Mon, 18 Jan 2021 17:05:55 +0000 (17:05 +0000)]
Added a comment

5 years agoupdate
Joey Hess [Mon, 18 Jan 2021 17:02:28 +0000 (13:02 -0400)]
update

5 years agocomment
Joey Hess [Mon, 18 Jan 2021 16:42:34 +0000 (12:42 -0400)]
comment

5 years agoBug fix: Fix tilde expansion in ssh urls when the tilde is the last character in...
Joey Hess [Mon, 18 Jan 2021 16:22:48 +0000 (12:22 -0400)]
Bug fix: Fix tilde expansion in ssh urls when the tilde is the last character in the url. Thanks, Grond for the patch.

5 years agoMerge branch 'master' of ssh://git-annex.branchable.com
Joey Hess [Mon, 18 Jan 2021 16:18:23 +0000 (12:18 -0400)]
Merge branch 'master' of ssh://git-annex.branchable.com

5 years agoinitial report on prop_relPathDirToFileAbs_basics windows FAIL
yarikoptic [Mon, 18 Jan 2021 15:50:24 +0000 (15:50 +0000)]
initial report on prop_relPathDirToFileAbs_basics windows FAIL

5 years agoAdded a comment
Lukey [Mon, 18 Jan 2021 14:31:06 +0000 (14:31 +0000)]
Added a comment

5 years agoAdded a comment
Lukey [Mon, 18 Jan 2021 14:11:45 +0000 (14:11 +0000)]
Added a comment

5 years ago(no commit message)
Lukey [Mon, 18 Jan 2021 14:02:19 +0000 (14:02 +0000)]

5 years agoAdded a comment: Works fine after changing mount point
seanl@fe5df935169a5440a52bdbfc5fece85cdd002d68 [Sun, 17 Jan 2021 07:24:38 +0000 (07:24 +0000)]
Added a comment: Works fine after changing mount point

5 years ago(no commit message)
seanl@fe5df935169a5440a52bdbfc5fece85cdd002d68 [Sun, 17 Jan 2021 06:36:20 +0000 (06:36 +0000)]

5 years agoimprove
Joey Hess [Fri, 15 Jan 2021 15:31:08 +0000 (11:31 -0400)]
improve

5 years ago(no commit message)
grond66@79ca29ba964cd0d8e2f352871d54452e4a9dad88 [Fri, 15 Jan 2021 13:07:28 +0000 (13:07 +0000)]

5 years ago(no commit message)
grond66@79ca29ba964cd0d8e2f352871d54452e4a9dad88 [Fri, 15 Jan 2021 13:01:50 +0000 (13:01 +0000)]

5 years agoAdded a comment
mih [Fri, 15 Jan 2021 08:32:58 +0000 (08:32 +0000)]
Added a comment

5 years ago(no commit message)
grond66@79ca29ba964cd0d8e2f352871d54452e4a9dad88 [Fri, 15 Jan 2021 02:24:25 +0000 (02:24 +0000)]

5 years agoremoved
eric.w@eee65cd362d995ced72640c7cfae388ae93a4234 [Thu, 14 Jan 2021 22:16:20 +0000 (22:16 +0000)]
removed

5 years agoremoved
eric.w@eee65cd362d995ced72640c7cfae388ae93a4234 [Thu, 14 Jan 2021 22:15:54 +0000 (22:15 +0000)]
removed

5 years agoAdded a comment
eric.w@eee65cd362d995ced72640c7cfae388ae93a4234 [Thu, 14 Jan 2021 22:15:12 +0000 (22:15 +0000)]
Added a comment

5 years agoclose as not a git-annex bug
Joey Hess [Wed, 13 Jan 2021 18:51:44 +0000 (14:51 -0400)]
close as not a git-annex bug

5 years agoBug fix: export with -J could fail when two files had the same content.
Joey Hess [Wed, 13 Jan 2021 18:50:48 +0000 (14:50 -0400)]
Bug fix: export with -J could fail when two files had the same content.

Exporting is done inside a call to writeLockDbWhile which guarantees there
is only one process uploading to a given ExportLocation.

5 years agofix build warning
Joey Hess [Wed, 13 Jan 2021 18:48:41 +0000 (14:48 -0400)]
fix build warning

5 years agoWindows: Work around win32 length limits when dealing with lock files
Joey Hess [Wed, 13 Jan 2021 18:38:35 +0000 (14:38 -0400)]
Windows: Work around win32 length limits when dealing with lock files

5 years agoexport TestableFilePath constructor
Joey Hess [Wed, 13 Jan 2021 17:23:35 +0000 (13:23 -0400)]
export TestableFilePath constructor

Useful for eg, replicating failures in ghci. No need for this to be a
smart constructor, as long as it's used with valid filepaths, it's ok
and if not the test breaks.

5 years agorewrite prop_relPathDirToFileAbs_basics
Joey Hess [Wed, 13 Jan 2021 17:21:15 +0000 (13:21 -0400)]
rewrite prop_relPathDirToFileAbs_basics

This was not a good test, it broke the requirement that
relPathDirToFileAbs take absolute paths. And it failed when the two
input paths were eg, the same but differently normalized.

Replaced with some tests of the real basics of that function.

5 years agoclose
Joey Hess [Wed, 13 Jan 2021 17:19:54 +0000 (13:19 -0400)]
close

5 years agocomment
Joey Hess [Wed, 13 Jan 2021 16:58:21 +0000 (12:58 -0400)]
comment