Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
|
|
While gathering the list of _really_ changed files, record the name they
used to have such that we can deal with renames.
Git does not really support renames, and calculates this at runtime also
providing a 'score' indicating the likelyhood of being a rename. We
currently only treat as renames those situations with a score > 75 (out
of 100).
|
|
If files are added then removed, or changed and then reset, don't show
them in the webrev.
We do this by moving the call to whatchanged into the perl itself, and
then effectively doing a join on filename between the 'diff' file list,
and the whatchanged file list.
|
|
As with mercurial, display the full set of checkin comments, rudely
concatenated, rather than just the first 'subject' comment line that
normal people would use, but Illumos steadfastly refuses to.
|
|
We made git function fundamentally differently, in terms of the webrev
"parent". Document (to the best degree I can).
- That the parent is the branch this branch is tracking, or
origin/master
- That a specified parent is anything at all that git accepts as a
tree, including reflog entries, etc, etc.
|
|
|
|
|
|
All these years, and still making that mistake...
|
|
In keeping with the rest of our support, there's no concept of "parent
workspace" so this is really just a matter of choosing where to put the
protodefs.
In passing fix the classic bug where GNU dirname of no arguments is an
error, and the rest of the script cries.
|
|
Similar to Mercurial support, list all files in the repository (which
is fast), then pair it down.
|
|
Reviewed by: Richard Lowe <richlowe@richlowe.net>
Reviewed by: Garrett D'Amore <garrett@nexenta.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
|
|
|
|
|
|
Since the git build_old_new changes directory, we need to store absolute
paths for the parent webrev support to function correctly.
|
|
build_old_new_* sadly need to be aware if the parent "workspace" is
actually a webrev, so they can get the old version of the file with cp
rather than whatever native means.
|
|
Adjust the way we build the file list, so that we don't trip over
revisions containing no files.
We now build a list of revision --> message mappings, and then get the
file lists for each revision, which is inefficient but otherwise does
the right things.
|
|
|
|
Make the git webrev act like the others, and get the 'new' side of the diff
from the working copy, not the most recent commit.
|
|
We previously processed files in whichever order the nawk associative
array stored its keys, which is hard to read. Instead, do so sorted by
file name.
|
|
A couple of places lost their use of $GIT, restore it
|
|
|
|
It seems most logical for webrev to default to comparing against the
branch from which 'git pull' would pull, only falling back to
'origin/master' if we can't figure that out.
|
|
It seems git ls-tree uses somewhat CWD-relative paths, and doesn't work at all
when in a directory not under git control)
|
|
Reviewed by: Dan McDonald <danmcd@nexenta.com>
Reviewed by: Olga Kryzhanoska <olga.kryzhanovska@gmail.com>
Approved by: Garrett D'Amore <garrett@nexenta.com>
|
|
Reviewed by: roland.mainz@nexenta.com
Reviewed by: garrett@nexenta.com
Approved by: garrett@nexenta.com
|
|
Reviewed by: Robert Gordon <rbg@openrbg.com>
Reviewed by: Gordon Ross <gwr@nexenta.com>
Reviewed by: George Wilson <gwilson@zfsmail.com>
Approved by: Garret D'Amore <garrett@nexenta.com>
|
|
Reviewed by: matt@greenviolet.net
Reviewed by: alan.coopersmith@gmail.com
Reviewed by: jason.brian.king@gmail.com
Approved by: garrett@nexenta.com
|
|
Reviewed by: Gordon Ross <gordon.w.ross@gmail.com>
Reviewed by: Albert Lee <trisk@opensolaris.org>
Approved by: Garrett D'Amore <garrett@nexenta.com>
|
|
Reviewed by: Rich Lowe <richlowe@richlowe.net>
Reviewed by: Albert Lee <trisk@nexenta.com>
Approved by: Gordon Ross <gwr@nexenta.com>
|
|
|
|
|
|
|
|
|
|
|
|
Reviewed by: garrett@nexenta.com
Reviewed by: matt@greenviolet.net
Approved by: garrett@nexenta.com
|
|
Bare minimum documentation of the support for Git
- How we find the workspace root
- How we (don't) determine the parent workspace
|
|
If the parent branch name includes a '/', take the first part to be an origin
name, translate it to a URI (using the first found), and include it in the
pretty parent header, so people can more usefully know what a particular ref
name actually means.
|
|
Rearrange the code to add PRETTY_CWS and PRETY_PWS, set by each SCM supporting
it to a string to be used as the "pretty" name of the current and parent
workspace, respectively. At present this is used to avoid duplication of the
"(at <node>)" logic for Mercurial and Git.
|
|
The parent (webrev -p) can now be any commit-shaped thing (see
git-rev-parse(1)), it cannot, however, be the conventional filesystem path or
URL.
|
|
Remove some hanging references to git-active in uncommon use cases (an
explicitly specified, non-wx format file list), and some references in
commentary.
|
|
When building the old and new directories for git, check that the new
path exists before trying to chmod it -- in the case of files that were
deleted, it won't.
|
|
Remove git-active, and build the filelist within webrev itself. Unlike
with Mercurial, git-active was purely doing text-processing that could
be done easily, and tidily, within webrev itself.
|
|
|
|
|
|
Reviewed by: johnsonnenschein@gmail.com, garrett@nexenta.com
Approved by: garrett@nexenta.com
--HG--
extra : rebase_source : 8c86d0860bfe3facd8bffa7ccdf7988b92945dd9
|
|
303 mdb's tigen should use terminfo database from proto
Reviewed by: richlowe@richlowe.net
Reviewed by: garrett@nexenta.com
Approved by: garrett@nexenta.com
|
|
Reviewed by: garrett@nexenta.com
Approved by: gwr@nexenta.com
--HG--
extra : rebase_source : 226d476ba44d6290017d6774aab7ea3b07dce615
|
|
|
|
With all of the other SCMs supported by webrev, CODEMGR_WS, CWS,
etc. refer to the root of the workspace, not the root of the SCM
metadata directory. Make git follow this pattern, and append '.../.git'
as necessary
|