]> git.proxmox.com Git - libgit2.git/log
libgit2.git
9 years agoMerge pull request #3030 from linquize/symlink_supported
Edward Thomson [Fri, 10 Apr 2015 09:33:14 +0000 (11:33 +0200)]
Merge pull request #3030 from linquize/symlink_supported

If work_dir is not specified, use repo_dir to test if symlink is supported

9 years agoMerge branch 'pr/3035'
Edward Thomson [Fri, 10 Apr 2015 08:50:01 +0000 (10:50 +0200)]
Merge branch 'pr/3035'

9 years agoMerge pull request #3038 from pks-t/fix-regcomp-retval-check
Edward Thomson [Fri, 10 Apr 2015 08:32:42 +0000 (10:32 +0200)]
Merge pull request #3038 from pks-t/fix-regcomp-retval-check

Fix checking of return value for regcomp.

9 years agoFix checking of return value for regcomp.
Patrick Steinhardt [Fri, 10 Apr 2015 07:36:38 +0000 (09:36 +0200)]
Fix checking of return value for regcomp.

The regcomp function returns a non-zero value if compilation of
a regular expression fails. In most places we only check for
negative values, but positive values indicate an error, as well.
Fix this tree-wide, fixing a segmentation fault when calling
git_config_iterator_glob_new with an invalid regexp.

9 years agoMerge pull request #3033 from pks-t/describe-oid-fallback
Edward Thomson [Wed, 8 Apr 2015 14:32:28 +0000 (16:32 +0200)]
Merge pull request #3033 from pks-t/describe-oid-fallback

Fix describe without refs and OID fallbacks

9 years agoISO C90 forbids mixed declarations and code
Jozef Matula [Wed, 8 Apr 2015 14:30:44 +0000 (16:30 +0200)]
ISO C90 forbids mixed declarations and code

GIT_UNUSED() macro generates code therefore it should be used after
variable declarations.

9 years agodescribe: only abort without tags if fallback is not allowed.
Patrick Steinhardt [Tue, 7 Apr 2015 12:44:08 +0000 (14:44 +0200)]
describe: only abort without tags if fallback is not allowed.

When no reference names could be found we did error out when trying to describe
a commit. This is wrong, though, when the option to fall back to a commit's
object ID is set.

9 years agodescribe example: enable building by default.
Patrick Steinhardt [Tue, 7 Apr 2015 12:43:04 +0000 (14:43 +0200)]
describe example: enable building by default.

9 years agoTest: Create repo in while current dir is readonly and checkout symlink
Linquize [Fri, 3 Apr 2015 16:29:01 +0000 (00:29 +0800)]
Test: Create repo in while current dir is readonly and checkout symlink

9 years agoFor bare repository, use repo_dir to test if symlinks are supported
Linquize [Fri, 3 Apr 2015 11:01:20 +0000 (19:01 +0800)]
For bare repository, use repo_dir to test if symlinks are supported

9 years agoMerge pull request #3031 from git-up/git_index_add
Edward Thomson [Sat, 4 Apr 2015 14:17:59 +0000 (09:17 -0500)]
Merge pull request #3031 from git-up/git_index_add

Entry argument passed to git_index_add_frombuffer() should be const

9 years agoEntry argument passed to git_index_add_frombuffer() should be const
Pierre-Olivier Latour [Sat, 4 Apr 2015 01:59:11 +0000 (18:59 -0700)]
Entry argument passed to git_index_add_frombuffer() should be const

9 years agoMerge pull request #3022 from ethomson/gitignore_vim
Edward Thomson [Wed, 1 Apr 2015 15:13:04 +0000 (11:13 -0400)]
Merge pull request #3022 from ethomson/gitignore_vim

gitignore: ignore vim swapfiles

9 years agogitignore: ignore vim swapfiles
Edward Thomson [Mon, 30 Mar 2015 13:33:02 +0000 (09:33 -0400)]
gitignore: ignore vim swapfiles

9 years agoMerge pull request #3011 from ethomson/filter_zero_tempbuf
Edward Thomson [Wed, 25 Mar 2015 17:14:28 +0000 (13:14 -0400)]
Merge pull request #3011 from ethomson/filter_zero_tempbuf

Clear temporary buffer when filtering

9 years agofilter: clear the temp_buf if we're using one
Edward Thomson [Mon, 23 Mar 2015 17:12:55 +0000 (13:12 -0400)]
filter: clear the temp_buf if we're using one

If we are using a temporary buffer for filtering, be sure to clear
it before using it, in case the file that we are filtering is empty.

9 years agoIllustrate bad checkout on Windows
Jacques Germishuys [Sun, 22 Mar 2015 17:06:53 +0000 (19:06 +0200)]
Illustrate bad checkout on Windows

9 years agoMerge pull request #3017 from ethomson/leaks
Carlos Martín Nieto [Wed, 25 Mar 2015 01:22:42 +0000 (02:22 +0100)]
Merge pull request #3017 from ethomson/leaks

squash some leaks

9 years agosquash some leaks
Edward Thomson [Tue, 24 Mar 2015 20:58:00 +0000 (20:58 +0000)]
squash some leaks

9 years agoMerge pull request #2990 from leoyanggit/custom_param
Edward Thomson [Tue, 24 Mar 2015 18:03:51 +0000 (14:03 -0400)]
Merge pull request #2990 from leoyanggit/custom_param

Add a custom param to git_smart_subtransport_definition

9 years agoMerge pull request #2986 from tkelman/mingw_winhttp
Edward Thomson [Tue, 24 Mar 2015 13:06:20 +0000 (09:06 -0400)]
Merge pull request #2986 from tkelman/mingw_winhttp

WinHTTP for MinGW

9 years agoMerge pull request #2947 from libgit2/cmn/notes-buf
Edward Thomson [Tue, 24 Mar 2015 12:34:12 +0000 (08:34 -0400)]
Merge pull request #2947 from libgit2/cmn/notes-buf

note: use a git_buf to return the default namespace

9 years agoMerge pull request #2858 from pks-t/remote-squashed
Carlos Martín Nieto [Tue, 24 Mar 2015 09:37:12 +0000 (10:37 +0100)]
Merge pull request #2858 from pks-t/remote-squashed

examples: add remote example.

9 years agoMerge pull request #3010 from ethomson/checkout_invalid_path_err
Carlos Martín Nieto [Tue, 24 Mar 2015 09:18:44 +0000 (10:18 +0100)]
Merge pull request #3010 from ethomson/checkout_invalid_path_err

checkout: report correct invalid path

9 years agocheckout: report correct invalid path
Edward Thomson [Mon, 23 Mar 2015 15:30:30 +0000 (11:30 -0400)]
checkout: report correct invalid path

9 years agogit: make sure to close the network stream
Carlos Martín Nieto [Sat, 21 Mar 2015 20:48:03 +0000 (21:48 +0100)]
git: make sure to close the network stream

In case of a bad url or other error during the connection setup, we
close the stream via free.

9 years agoMerge pull request #3003 from shawnl/master
Edward Thomson [Fri, 20 Mar 2015 20:57:32 +0000 (16:57 -0400)]
Merge pull request #3003 from shawnl/master

fix changelog

9 years agofix changelog
Shawn Landden [Fri, 20 Mar 2015 17:18:12 +0000 (10:18 -0700)]
fix changelog

9 years agoMerge pull request #3000 from libgit2/vmg/mkdir-ext
Edward Thomson [Thu, 19 Mar 2015 19:10:58 +0000 (15:10 -0400)]
Merge pull request #3000 from libgit2/vmg/mkdir-ext

mkdir-ext: Assume directories don't exist; fix all race cases

9 years agomkdir-ext: Retry lstat on EEXIST race
Vicent Marti [Thu, 19 Mar 2015 17:38:13 +0000 (18:38 +0100)]
mkdir-ext: Retry lstat on EEXIST race

9 years agoUse swprintf_s everywhere except mingw.org
Tony Kelman [Thu, 19 Mar 2015 14:25:25 +0000 (07:25 -0700)]
Use swprintf_s everywhere except mingw.org

9 years agoexamples: add remote example.
Patrick Steinhardt [Mon, 26 Jan 2015 14:36:34 +0000 (15:36 +0100)]
examples: add remote example.

9 years agoMerge pull request #2996 from ethomson/dll_comments
Carlos Martín Nieto [Thu, 19 Mar 2015 00:03:41 +0000 (01:03 +0100)]
Merge pull request #2996 from ethomson/dll_comments

win32 resource: allow custom comments field in DLL

9 years agowin32 resource: allow custom comments field in DLL
Edward Thomson [Wed, 18 Mar 2015 22:25:14 +0000 (18:25 -0400)]
win32 resource: allow custom comments field in DLL

9 years agoMerge pull request #2967 from jacquesg/merge-whitespace
Edward Thomson [Wed, 18 Mar 2015 17:17:04 +0000 (13:17 -0400)]
Merge pull request #2967 from jacquesg/merge-whitespace

Allow merges of files (and trees) with whitespace problems/fixes

9 years agoAdd a custom param to git_smart_subtransport_definition
Leo Yang [Tue, 17 Mar 2015 16:49:33 +0000 (12:49 -0400)]
Add a custom param to git_smart_subtransport_definition

The smart transport has already take the payload param. For the
sub transport a payload param is useful for the implementer.

9 years agoMerge pull request #2993 from ethomson/clar_update
Carlos Martín Nieto [Wed, 18 Mar 2015 04:04:50 +0000 (05:04 +0100)]
Merge pull request #2993 from ethomson/clar_update

clar: update to 08f434d

9 years agoMerge pull request #2955 from git-up/update_tips_fixes
Carlos Martín Nieto [Wed, 18 Mar 2015 04:04:12 +0000 (05:04 +0100)]
Merge pull request #2955 from git-up/update_tips_fixes

Update tips fixes

9 years agoMerge remote-tracking branch 'ethomson/submodule_8dot3'
Carlos Martín Nieto [Wed, 18 Mar 2015 03:59:16 +0000 (04:59 +0100)]
Merge remote-tracking branch 'ethomson/submodule_8dot3'

9 years agoMerge pull request #2985 from claudiuolteanu/patch-1
Edward Thomson [Wed, 18 Mar 2015 03:35:52 +0000 (23:35 -0400)]
Merge pull request #2985 from claudiuolteanu/patch-1

Lower case the include directive of windows header

9 years agoMerge pull request #2983 from jeffhostetler/jeffhostetler/perf_merge_lazy_binary_check
Edward Thomson [Wed, 18 Mar 2015 02:22:38 +0000 (22:22 -0400)]
Merge pull request #2983 from jeffhostetler/jeffhostetler/perf_merge_lazy_binary_check

PERF: In MERGE, lazily compute is_binary

9 years agoclar: update to 08f434d
Edward Thomson [Tue, 17 Mar 2015 23:04:48 +0000 (19:04 -0400)]
clar: update to 08f434d

9 years agoMerge pull request #2991 from sba1/fix-uninitialized-data-access-remote-update-tips
Edward Thomson [Tue, 17 Mar 2015 22:35:28 +0000 (18:35 -0400)]
Merge pull request #2991 from sba1/fix-uninitialized-data-access-remote-update-tips

Initialize refs vector in git_remote_update_tips().

9 years agoInitialize refs vector in git_remote_update_tips().
Sebastian Bauer [Tue, 17 Mar 2015 20:38:58 +0000 (21:38 +0100)]
Initialize refs vector in git_remote_update_tips().

Otherwise, bailing out early when ls_to_vector() fails accesses
uninitialized memory.

9 years agonote: use a git_buf to return the default namespace
Carlos Martín Nieto [Wed, 4 Mar 2015 00:23:20 +0000 (01:23 +0100)]
note: use a git_buf to return the default namespace

The caller has otherwise no way to know how long the string will be
allocated or ability to free it.

This fixes #2944.

9 years agoset PKG_CONFIG_LIBDIR in toolchain-mingw32.cmake
Tony Kelman [Tue, 17 Mar 2015 19:05:07 +0000 (12:05 -0700)]
set PKG_CONFIG_LIBDIR in toolchain-mingw32.cmake

9 years agoAdded changelog entries for whitespace and patience merges.
Jacques Germishuys [Tue, 17 Mar 2015 14:34:36 +0000 (16:34 +0200)]
Added changelog entries for whitespace and patience merges.

9 years agoMerge pull request #2962 from libgit2/cmn/reflog-annotated
Edward Thomson [Tue, 17 Mar 2015 14:21:28 +0000 (10:21 -0400)]
Merge pull request #2962 from libgit2/cmn/reflog-annotated

Add annotated versions of ref-modying functions

9 years agoMerge pull request #2972 from libgit2/cmn/pack-objects-walk
Edward Thomson [Tue, 17 Mar 2015 14:06:50 +0000 (10:06 -0400)]
Merge pull request #2972 from libgit2/cmn/pack-objects-walk

[WIP] Smarter pack-building

9 years agoMerge pull request #2982 from libgit2/cmn/stream-check-ec
Edward Thomson [Tue, 17 Mar 2015 13:44:56 +0000 (09:44 -0400)]
Merge pull request #2982 from libgit2/cmn/stream-check-ec

Don't ask for a stream's certificate unless it's encrypted

9 years agoMerge pull request #2977 from pks-t/submodule-sync-url-fix
Edward Thomson [Tue, 17 Mar 2015 13:44:25 +0000 (09:44 -0400)]
Merge pull request #2977 from pks-t/submodule-sync-url-fix

Fix git_submodule_sync writing URL to wrong key.

9 years agouse a different .def file for 64 bit
Tony Kelman [Mon, 16 Mar 2015 21:49:23 +0000 (14:49 -0700)]
use a different .def file for 64 bit

9 years agoLower case the include directive of windows header
Claudiu Olteanu [Mon, 16 Mar 2015 21:08:16 +0000 (23:08 +0200)]
Lower case the include directive of windows header

Since the Linux platform has a case sensitive file system, the header name should be lower case for cross compiling purposes. (On Linux, the mingw header is called ```windows.h```).

9 years agoFix out-of-tree build
Tony Kelman [Mon, 16 Mar 2015 20:30:10 +0000 (13:30 -0700)]
Fix out-of-tree build

link against crypt32 for CertFreeCertificateContext

9 years agoFix build on mingw-w64
Alex Crichton [Tue, 26 Aug 2014 18:18:33 +0000 (11:18 -0700)]
Fix build on mingw-w64

9 years agopatch so mingw-w64 can build
klutzy [Mon, 16 Mar 2015 19:46:33 +0000 (12:46 -0700)]
patch so mingw-w64 can build

9 years agoWin32: Enable WinHTTP for MinGW
Philip Kelley [Mon, 9 Jun 2014 15:43:25 +0000 (11:43 -0400)]
Win32: Enable WinHTTP for MinGW

9 years agoPERF: In MERGE, lazily compute is_binary
Jeff Hostetler [Mon, 16 Mar 2015 19:54:53 +0000 (15:54 -0400)]
PERF: In MERGE, lazily compute is_binary

9 years agohttp: do not try to use the cert callback on unencrypted streams
Carlos Martín Nieto [Mon, 16 Mar 2015 18:41:50 +0000 (19:41 +0100)]
http: do not try to use the cert callback on unencrypted streams

When the user has a certificate check callback set, we still have to
check whether the stream we're using is even capable of providing a
certificate.

In the case of an unencrypted certificate, do not ask for it from the
stream, and do not call the callback.

9 years agoAdd failing test for cert callback with non-ecrypted stream
Carlos Martín Nieto [Mon, 16 Mar 2015 18:39:05 +0000 (19:39 +0100)]
Add failing test for cert callback with non-ecrypted stream

When we have an HTTP stream and have set the certificatre check
callback, we currently fail as we ask the unencrypted stream for its
certificate.

9 years agoCHANGELOG: add notes about the annotated ref-modifying functions
Carlos Martín Nieto [Sat, 7 Mar 2015 11:09:41 +0000 (12:09 +0100)]
CHANGELOG: add notes about the annotated ref-modifying functions

9 years agoAdd tests for the annotated versions of ref-modifying functions
Carlos Martín Nieto [Fri, 6 Mar 2015 23:30:40 +0000 (00:30 +0100)]
Add tests for the annotated versions of ref-modifying functions

This also brings the soft-reset tests back to life. The function name
was missing an underscore, meaning they had not been running.

9 years agoannotated_commit: provide a constructor from a revspec
Carlos Martín Nieto [Fri, 6 Mar 2015 23:06:02 +0000 (00:06 +0100)]
annotated_commit: provide a constructor from a revspec

This extra constructor will be useful for the annotated versions of
ref-modifying functions, as it allows us to create a commit with the
extended sha syntax which was used to retrieve it.

9 years agoAdd annotated commit versions of reflog-modifying functions
Carlos Martín Nieto [Fri, 6 Mar 2015 22:51:40 +0000 (23:51 +0100)]
Add annotated commit versions of reflog-modifying functions

We do not always want to put the id directly into the reflog, but we
want to speicfy what a user typed. For this use-case we provide
annotated version of a few functions which let the caller specify what
user-friendly name was used when asking for the operation.

9 years agoAdded options to enable patience and minimal diff drivers
Jacques Germishuys [Thu, 12 Mar 2015 11:16:09 +0000 (13:16 +0200)]
Added options to enable patience and minimal diff drivers

9 years agoCollapse whitespace flags into git_merge_file_flags_t
Jacques Germishuys [Thu, 12 Mar 2015 10:36:09 +0000 (12:36 +0200)]
Collapse whitespace flags into git_merge_file_flags_t

9 years agoRenamed git_merge_options 'flags' to 'tree_flags'
Jacques Germishuys [Thu, 12 Mar 2015 10:29:47 +0000 (12:29 +0200)]
Renamed git_merge_options 'flags' to 'tree_flags'

9 years agoAdded tests to merge files and branches with whitespace problems and fixes
Jacques Germishuys [Mon, 9 Mar 2015 15:03:03 +0000 (17:03 +0200)]
Added tests to merge files and branches with whitespace problems and fixes

9 years agoAllow for merges with whitespace discrepancies
Jacques Germishuys [Mon, 9 Mar 2015 15:02:52 +0000 (17:02 +0200)]
Allow for merges with whitespace discrepancies

9 years agoMerge pull request #2975 from git-up/diff_index_retain
Carlos Martín Nieto [Sat, 14 Mar 2015 23:24:36 +0000 (00:24 +0100)]
Merge pull request #2975 from git-up/diff_index_retain

Avoid retaining / releasing the index more than necessary when GIT_DIFF_...

9 years agoindexer: bring back the error message on duplcate commits
Carlos Martín Nieto [Fri, 13 Mar 2015 18:41:40 +0000 (19:41 +0100)]
indexer: bring back the error message on duplcate commits

It turns out that erroring out on duplicate commits is the right thing
to do, but git was not hitting the bug on the server-side.

Bring back a descriptive error message in case of duplicate entries and
error out.

9 years agoindexer: don't worry about duplicate objects
Carlos Martín Nieto [Fri, 13 Mar 2015 17:28:07 +0000 (18:28 +0100)]
indexer: don't worry about duplicate objects

If a packfile includes duplicate objects, we can choose to use the
secon copy instead of the first by using the same logic as if it were
the first.

Change the error condition from 0 to -1, which indicates a bad resize,
and set the OOM message in that case.

This does mean we will leak the first copy of the object. We can deal
with that later, but making fetches work is more important.

9 years agoindexer: set an error message on duplicate objects in pack
Carlos Martín Nieto [Fri, 13 Mar 2015 17:00:15 +0000 (18:00 +0100)]
indexer: set an error message on duplicate objects in pack

While this is not even close to a fix, we can at least set an error
message so we know which error we are facing. Up to know we just
returned an error without a message.

9 years agoPut back the number of expected references to 6 from the test repo
Carlos Martín Nieto [Fri, 13 Mar 2015 16:52:07 +0000 (17:52 +0100)]
Put back the number of expected references to 6 from the test repo

This was but down to 5 when GitHub made a change to their server which
made them stop honouring the include-tag request.

This has recently been corrected, so we can bring it back up to six.

9 years agoMerge pull request #2973 from libgit2/cmn/local-pack-threading
Edward Thomson [Thu, 12 Mar 2015 22:04:25 +0000 (18:04 -0400)]
Merge pull request #2973 from libgit2/cmn/local-pack-threading

local: create pack with multiple threads

9 years agoFix git_submodule_sync writing URL to wrong key.
Patrick Steinhardt [Thu, 12 Mar 2015 15:05:07 +0000 (16:05 +0100)]
Fix git_submodule_sync writing URL to wrong key.

Currently git_submodule_sync writes the submodule's URL to the
key 'branch.<REMOTE_NAME>.remote' while the reference
implementation of `git submodule sync` writes to
'remote.<REMOTE_NAME>.url', which is the intended behavior
according to git-submodule(1).

9 years agoAvoid retaining / releasing the index more than necessary when GIT_DIFF_UPDATE_INDEX...
Pierre-Olivier Latour [Thu, 12 Mar 2015 02:29:36 +0000 (19:29 -0700)]
Avoid retaining / releasing the index more than necessary when GIT_DIFF_UPDATE_INDEX is enabled

9 years agoFixed active refspecs not reset by git_remote_upload()
Pierre-Olivier Latour [Fri, 6 Mar 2015 07:06:31 +0000 (23:06 -0800)]
Fixed active refspecs not reset by git_remote_upload()

9 years agoFixed update_tips callback called for failed pushed references
Pierre-Olivier Latour [Fri, 6 Mar 2015 06:13:54 +0000 (22:13 -0800)]
Fixed update_tips callback called for failed pushed references

The current implementation does not set 'fire_callback' back to 0 for failed updates so the callback still fires.

Instead of adding yet another condition check to set 'fire_callback' to 0 if needed, considering this function should be a no-op for failed updates anyway, the best fix is to simplify its logic to check upfront if the update is a failed one.

9 years agoMerge commit 'refs/pull/2879/head' of ssh://github.com/libgit2/libgit2
Carlos Martín Nieto [Wed, 11 Mar 2015 16:55:39 +0000 (17:55 +0100)]
Merge commit 'refs/pull/2879/head' of ssh://github.com/libgit2/libgit2

9 years agolocal: create pack with multiple threads
Carlos Martín Nieto [Wed, 11 Mar 2015 16:16:39 +0000 (17:16 +0100)]
local: create pack with multiple threads

The default behaviour for the packbuilder is to perform the work in a
single thread, which is fine for the public API, but we currently have
no way for a user to determine the number of threads to use when
creating the packfile, which makes our clone behaviour over the
filesystem quite a bit slower than what git offers.

This is a very particular scenario, in which we avoid spawning git by
being ourselves the server-side, so it's probably ok to auto-set the
threading, as the upload-pack process would do if we were talking to
git.

9 years agoMerge pull request #2969 from adrienthebo/remote_add_invalid_refspec
Carlos Martín Nieto [Wed, 11 Mar 2015 15:13:31 +0000 (16:13 +0100)]
Merge pull request #2969 from adrienthebo/remote_add_invalid_refspec

refspec: report errors when parsing an invalid refspec

9 years agoMerge pull request #2970 from ethomson/inmemory_bare
Carlos Martín Nieto [Wed, 11 Mar 2015 02:03:54 +0000 (03:03 +0100)]
Merge pull request #2970 from ethomson/inmemory_bare

"In-memory" repos are bare by default

9 years agolocal: let the packbuilder perform smarter object insertion
Carlos Martín Nieto [Wed, 4 Mar 2015 19:57:52 +0000 (20:57 +0100)]
local: let the packbuilder perform smarter object insertion

Currently we use the most naïve and inefficient method for figuring out
which objects to send to the remote whereby we end up trying to insert
subdirs which have not changed multiple times.

Instead, make use of the packbuilder's built-in more efficient method
which uses the walk to feed the object list and avoids inserting an
object and its descendants.

9 years agopack-objects: fill a packbuilder from a walk
Carlos Martín Nieto [Sat, 11 Oct 2014 13:48:29 +0000 (15:48 +0200)]
pack-objects: fill a packbuilder from a walk

Most use-cases for the object packer communicate in terms of commits
which each side has. We already have an object to specify this
relationship between commits, namely git_revwalk.

By knowing which commits we want to pack and which the other side
already has, we can perform similar optimisations to git, by marking
each tree as interesting or uninteresting only once, and not sending
those trees which we know the other side has.

9 years agoReorder some khash declarations
Carlos Martín Nieto [Sun, 12 Oct 2014 09:42:31 +0000 (11:42 +0200)]
Reorder some khash declarations

Keep the definitions in the headers, while putting the declarations in
the C files. Putting the function definitions in headers causes
them to be duplicated if you include two headers with them.

9 years agoMerge pull request #2971 from ethomson/revert_fastfail
Carlos Martín Nieto [Wed, 11 Mar 2015 01:14:22 +0000 (02:14 +0100)]
Merge pull request #2971 from ethomson/revert_fastfail

Revert "fastfail" from the mingw PR

9 years agorepository_new: test its bareness
Edward Thomson [Tue, 10 Mar 2015 20:49:24 +0000 (16:49 -0400)]
repository_new: test its bareness

9 years agorefspec: set err message on invalid refspec
Adrien Thebo [Wed, 4 Mar 2015 21:29:13 +0000 (13:29 -0800)]
refspec: set err message on invalid refspec

If a refspec could not be parsed, the git_refspec__parse function would
return an error value but would not provide additional error information
for the callers. This commit amends that function to set a more useful
error message.

9 years agoRevert "add fast-fail feature for pending builds in PR's"
Edward Thomson [Tue, 10 Mar 2015 20:08:44 +0000 (16:08 -0400)]
Revert "add fast-fail feature for pending builds in PR's"

This reverts commit 8008ab6a4158416d095212c9782fa1622ef994a5.

9 years agoRevert "Implement fail-fast for Travis as well"
Edward Thomson [Tue, 10 Mar 2015 20:08:34 +0000 (16:08 -0400)]
Revert "Implement fail-fast for Travis as well"

This reverts commit ba6c53b91b1d9c9dc49b7c0d8e8efbd9842c31dc.

9 years agorepository: in-memory repos are bare by default
Edward Thomson [Tue, 10 Mar 2015 19:02:02 +0000 (15:02 -0400)]
repository: in-memory repos are bare by default

9 years agoMerge pull request #2964 from git-up/fixes
Edward Thomson [Mon, 9 Mar 2015 14:18:39 +0000 (10:18 -0400)]
Merge pull request #2964 from git-up/fixes

Fixes

9 years agoMerge pull request #2946 from tkelman/appveyor-mingw
Edward Thomson [Mon, 9 Mar 2015 13:18:37 +0000 (09:18 -0400)]
Merge pull request #2946 from tkelman/appveyor-mingw

RFC: add mingw to appveyor matrix

9 years agoRemoved unnecessary GIT_CHECKOUT_SKIP_UNMERGED for GIT_RESET_HARD
Pierre-Olivier Latour [Sun, 8 Mar 2015 21:10:02 +0000 (14:10 -0700)]
Removed unnecessary GIT_CHECKOUT_SKIP_UNMERGED for GIT_RESET_HARD

9 years agoFixed build warning
Pierre-Olivier Latour [Sun, 8 Mar 2015 21:09:50 +0000 (14:09 -0700)]
Fixed build warning

9 years agoMerge pull request #2963 from libgit2/cmn/rename-reflog
Carlos Martín Nieto [Sun, 8 Mar 2015 18:58:51 +0000 (19:58 +0100)]
Merge pull request #2963 from libgit2/cmn/rename-reflog

refdb: use the same id for old and new when renaming a reference

9 years agorefdb: use the same id for old and new when renaming a reference
Carlos Martín Nieto [Sun, 8 Mar 2015 15:46:33 +0000 (16:46 +0100)]
refdb: use the same id for old and new when renaming a reference

When we rename a reference, we want the old and new ids to be the same
one (as we did not change it). The normal code path looks up the old id
from the current value of the brtanch, but by the time we look it up, it
does not exist anymore and thus we write a zero id.

Pass the old id explicitly instead.

9 years agoMerge pull request #2961 from ethomson/filter_relative_paths
Edward Thomson [Fri, 6 Mar 2015 21:52:07 +0000 (16:52 -0500)]
Merge pull request #2961 from ethomson/filter_relative_paths

Filter relative paths

9 years agoMerge pull request #2950 from libgit2/ntk/reflog_branch_create
Carlos Martín Nieto [Fri, 6 Mar 2015 21:44:15 +0000 (22:44 +0100)]
Merge pull request #2950 from libgit2/ntk/reflog_branch_create

branch: fix generated reflog message upon creation