4 The latest copy of this document is found at
6 http://kernel.org/git/?p=git/git.git;a=blob;hb=todo;f=TODO
9 What to expect from now on
10 ==========================
12 This is written in a form of to-do list for me, so if I say
13 "accept patch", it means I do not currently plan to do that
14 myself. People interested in seeing it materialize please take
15 a hint. Also whatever I marked "Perhaps" do not have to happen
16 if ever -- only if somebody cares enough and submits a clean
23 * No pending issues at the moment. "Revamp Tutorial" initiative
24 by Bruce Fields ongoing and things are looking better.
30 * Rehash "git commit" with various parameters to be more
31 intuitive without breaking traditinal users too much. We need
32 to phase this in, especially if we are going to change "git
33 commit" to imply the current "git commit -a" behaviour.
35 * "intent to add" index entries.
37 * Plug-in file-level merges. On the other hand, we may not even
38 need this; just tell people to run "xxdiff -U" on the working
41 * Doing a merge in a separate directory.
43 * Make 'format-patch' take revision limiters similar to
44 rev-list. For example:
47 ....---x---o---o---x---o---o
54 we should be able to format commits 'o', without duplicates,
57 $ git format-patch ^A ^B C
59 Currently the closest approximation is
61 $ git format-patch A..C B..C
63 which results in the last two commits including C formatted
70 * Libification. There are many places "run once" mentality is
71 ingrained in the management of basic data structures, which
72 need to be fixed. [Matthias Urlichs is already working on
73 this: <pan.2005.10.03.20.48.52.132570@smurf.noris.de>]
75 * Maybe a pack optimizer.
77 Given a set of objects and a set of refs (probably a handful
78 branch heads and point release tags), find a set of packs to
79 allow reasonably minimum download for all of these classes of
80 people: (1) somebody cloning the repository from scratch, (2)
81 somebody who tends to follow the master branch head reasonably
82 closely, (3) somebody who tends to follow only the point
85 This needs a matching smart on the dumb protocol downloader.
87 * Maybe an Emacs VC backend.
89 * Look at libified GNU diff CVS seems to use, or libxdiff.
90 [Daniel has his own diff tool almost ready to start
91 integrating and testing]
97 * Subprojects. I think the "bind commit" approach has been
98 outlined at sufficiently detailed level. Maybe find time to
99 actually start prototyping it?
101 <7vacdzkww3.fsf@assigned-by-dhcp.cox.net>
105 * Mark entries as "assume unchanged" in the index.
106 New option to update-index to set or drop the bit is needed.
108 - update-index --no-stat paths...
109 - update-index --with-stat paths...
111 Also a config item '[core] trust_stat = false' would enable
114 - "update-index" with or without --add would mark the path
115 after registering. Should we make the working tree file
116 read-only at this point?
118 - checkout-index -u would mark the path and makes the working
121 Impacts to various commands:
123 - update-index --refresh would ignore them.
125 - diff-files would say unchanged.
127 - diff-index without --cached acts the same way as diff-index
130 * Decide what to do about rebase applied to merged head. One
131 extreme is to allow rebase if "rev-list ours..theirs" gives
132 anything. This loosens the current merge-base based approach.
133 The other extreme is to refuse rebase if "rev-list
134 theirs..ours" contains any merge commit, which was discussed
137 <43CC695E.2020506@codeweavers.com>
139 * Decide what the right thing to do upon an empty merge commit,
140 when both branches happen to have obtained the same set of
141 changes through different history. Not recording such keeps
142 the history simpler, and the next merge would soon create a
143 true merge commit anyway, but this does not feel quite right.
145 <20060114021800.4688.qmail@web31803.mail.mud.yahoo.com>
147 * Perhaps a smarter HTTP anonymous download via CGI.
149 * Prepare to enable "always use symbolic refs for HEAD" patch.
150 We need a timeline to force Porcelains to get ready. All the
151 major ones should be ready now.
153 * diff stopping at the first output; qgit wants to know if this
154 tree has any A or D from the other tree and nothing else.
155 Would help internal tree-diff in rev-list as well.
157 * daemon --strict-symlink.
159 * daemon --no-user-dir, to make ~user still work with
160 --base-path. They ought to be independent.
162 * daemon --base-path does not apply automatically to whitelist
163 somehow feels wrong. If somebody cares enough, accept
166 * Perhaps detect cloning request in upload-pack and cache the
167 result for next cloning request until any of our refs change.
169 * Perhaps accept patch to optionally allow '--fuzz' in
170 'git-apply'. am/applymbox is _not_ the place to do it.
172 * Allow 'git apply' to accept GNU diff 2.7 output that forgets
173 to say '\No newline' if both input ends with incomplete
176 * Perhaps deal with "Files differ" (binary diff) in non C
179 * Maybe grok PGP signed text/plain in applymbox as well.
181 * Output full path in the "git-rev-list --objects" output, not
182 just the basename, and see the improved clustering results in
183 better packing [Tried, but did not work out well].
189 * Use parent info in 'diff-tree --stdin'.
191 * git-proxy should be spawned with sh -c 'command' $1 $2.
193 * test scripts for the relative directory path stuff.
195 * In a freshly created empty repository, `git fetch foo:bar`
196 works OK, but `git checkout bar` afterwards does not (missing