24
32
requires patching difflib to get it exactly right, or otherwise
25
33
calling out to GNU diff.
27
* Import ElementTree update patch.
29
* Syntax should be ``bzr export -r REV``.
31
* Plugins that provide commands. By just installing a file into some
32
directory (e.g. ``/usr/share/bzr/plugins``) it should be possible to
33
create new top-level commands (``bzr frob``). Extensions can be
34
written in either Python (in which case they use the bzrlib API) or
35
in a separate process (in sh, C, whatever). It should be possible
36
to get help for plugin commands.
38
* Smart rewrap text in help messages to fit in $COLUMNS (or equivalent
41
35
* -r option should take a revision-id as well as a revno.
43
* ``bzr info`` could show space used by working tree, versioned files,
44
unknown and ignored files.
46
37
* ``bzr info`` should count only people with distinct email addresses as
47
38
different committers. (Or perhaps only distinct userids?)
64
53
__ http://mail.python.org/pipermail/python-list/2001-April/037847.html
66
* Read and write locks on branch while it's open.
68
55
* Separate read and write version checks?
70
* ``bzr status FILE...``
72
* Check all commands have decent help.
74
* Autogenerate argument/option help.
57
* ``bzr status DIR`` should give status on all files under that
60
* ``bzr log DIR`` should give changes to any files within DIR.
76
62
* ``bzr inventory -r REV`` and perhaps unify this with ``bzr ls``,
77
63
giving options to display ids, types, etc.
79
* Atomic file class that renames into place when it's closed.
65
* Split BzrError into various more specific subclasses for different
66
errors people might want to catch.
68
* If the export destination ends in '.tar', '.tar.gz', etc then create
69
a tarball instead of a directory. (Need to actually make a
70
temporary directory and then tar that up.)
72
http://www.gelato.unsw.edu.au/archives/git/0504/2194.html
74
* RemoteBranch could maintain a cache either in memory or on disk. We
75
know more than an external cache might about which files are
76
immutable and which can vary. On the other hand, it's much simpler
77
to just use an external proxy cache.
79
Perhaps ~/.bzr/http-cache. Baz has a fairly simple cache under
80
~/.arch-cache, containing revision information encoded almost as a
81
bunch of archives. Perhaps we could simply store full paths.
83
* Maybe also store directories in the statcache so that we can quickly
84
identify that they still exist.
86
* Diff should show timestamps; for files from the working directory we
87
can use the file itself; for files from a revision we should use the
88
commit time of the revision.
90
* Perhaps split command infrastructure from the actual command
93
* Cleaner support for negative boolean options like --no-recurse.
95
* Statcache should possibly map all file paths to / separators
97
* quotefn doubles all backslashes on Windows; this is probably not the
98
best thing to do. What would be a better way to safely represent
99
filenames? Perhaps we could doublequote things containing spaces,
100
on the principle that filenames containing quotes are unlikely?
101
Nice for humans; less good for machine parsing.
103
* Patches should probably use only forward slashes, even on Windows,
104
otherwise Unix patch can't apply them. (?)
106
* Branch.update_revisions() inefficiently fetches revisions from the
107
remote server twice; once to find out what text and inventory they
108
need and then again to actually get the thing. This is a bit
111
One complicating factor here is that we don't really want to have
112
revisions present in the revision-store until all their constituent
113
parts are also stored.
115
The basic problem is that RemoteBranch.get_revision() and similar
116
methods return object, but what we really want is the raw XML, which
117
can be popped into our own store. That needs to be refactored.
119
* ``bzr status FOO`` where foo is ignored should say so.
121
* ``bzr mkdir A...`` should just create and add A.
123
* Guard against repeatedly merging any particular patch.
85
* Display command grammar in help messages rather than hardcoding it.
87
* Change command functions into Command() objects, like in hct, and
88
then the grammar can be described directly in there. Since all
89
option definitions are global we can define them just once and
90
reference them from each command.
92
* Selective commit of only some files.
96
Status should be handled differently because it needs to report on
97
deleted and unknown files. diff only needs to deal with versioned
100
* Merge Aaron's merge code.
102
130
* Merge revert patch.
104
* Turn on stat cache code, and add optimization about avoiding
105
dangerous cache entries.
107
132
* ``bzr mv`` that does either rename or move as in Unix.
109
* More efficient diff of only selected files.
134
* More efficient diff of only selected files. We should be able to
135
just get the id for the selected files, look up their location and
136
diff just those files. No need to traverse the entire inventories.
138
* ``bzr status DIR`` or ``bzr diff DIR`` should report on all changes
139
under that directory.
111
141
* Fix up Inventory objects to represent root object as an entry.
113
* Don't convert entire entry from
143
* Don't convert entire entry from ElementTree to an object when it is
144
read in, but rather wait until the program actually wants to know
115
147
* Extract changes from one revision to the next to a text form
116
148
suitable for transmission over email.
118
150
* More test cases.
152
- Selected-file commit
154
- Impossible selected-file commit: adding things in non-versioned
155
directories, crossing renames, etc.
120
157
* Write a reproducible benchmark, perhaps importing various kernel versions.
122
* Change test.sh from Bourne shell into something in pure Python so
123
that it can be more portable.
125
159
* Directly import diffs! It seems a bit redundant to need to rescan
126
160
the directory to work out what files diff added/deleted/changed when
127
161
all the information is there in the diff in the first place.
220
* More test framework:
222
- Class that describes the state of a working tree so we can just
225
* There are too many methods on Branch() that really manipulate the
226
WorkingTree. They should be moved across.
228
Also there are some methods which are duplicated on Tree and
229
Inventory objects, and it should be made more clear which ones are
230
proxies and which ones behave differently, and how.
232
* Try using XSLT to add some formatting to REST-generated HTML. Or
233
maybe write a small Python program that specifies a header and foot
234
for the pages and calls into the docutils libraries.
236
* --format=xml for log, status and other commands.
238
* Attempting to explicitly add a file that's already added should give
239
a warning; however there should be no warning for directories (since
240
we scan for new children) or files encountered in a directory that's
243
* Better handling of possible collisions on case-losing filesystems;
244
make sure a single file does not get added twice under different
247
* Clean up XML inventory:
249
- Use nesting rather than parent_id pointers.
251
- Hold the ElementTree in memory in the Inventory object and work
252
directly on that, rather than converting into Python objects every
253
time it is read in. Probably still exposoe it through some kind of
254
object interface though, but perhaps that should just be a proxy
257
- Less special cases for the root directory.
259
* Perhaps inventories should remember the revision in which each file
260
was last changed, as well as its current state? This is a bit
261
redundant but might often be interested to know.
263
* stat cache should perhaps only stat files as necessary, rather than
264
doing them all up-front. On the other hand, that disallows the
265
opimization of stating them in inode order.
267
* It'd be nice to pipeline multiple HTTP requests. Often we can
268
predict what will be wanted in future: all revisions, or all texts
269
in a particular revision, etc.
271
urlgrabber's docs say they are working on batched downloads; we
272
could perhaps ride on that or just create a background thread (ew).
274
* Paranoid mode where we never trust SHA-1 matches.
276
* Don't commit if there are no changes unless forced.
278
* --dry-run mode for commit? (Or maybe just run with
279
check-command=false?)
281
* Generally, be a bit more verbose unless --silent is specified.
283
* Function that finds all changes to files under a given directory;
284
perhaps log should use this if a directory is given.
286
* XML attributes might have trouble with filenames containing \n and
287
\r. Do we really want to support this? I think perhaps not.
289
* Remember execute bits, so that exports will work OK.
291
* Unify smart_add and plain Branch.add(); perhaps smart_add should
292
just build a list of files to add and pass that to the regular add
295
* Function to list a directory, saying in which revision each file was
296
last modified. Useful for web and gui interfaces, and slow to
297
compute one file at a time.
299
* unittest is standard, but the results are kind of ugly; would be
300
nice to make it cleaner.
302
* Check locking is correct during merge-related operations.
304
* Perhaps attempts to get locks should timeout after some period of
305
time, or at least display a progress message.
307
* Split out upgrade functionality from check command into a separate
310
* Don't pass around command classes but rather pass objects. This'd
311
make it cleaner to construct objects wrapping external commands.
313
* Track all merged-in revisions in a versioned add-only metafile.
216
347
files, and respecting selective commits. Run the pre-commit check
217
348
(e.g. compile and run test suite) in there.
350
Possibly this should be done by splitting the commit function into
351
several parts (under a single interface). It is already rather
352
large. Decomposition:
354
- find tree modifications and prepare in-memory inventory
356
- export that inventory to a temporary directory
358
- run the test in that temporary directory
360
- if that succeeded, continue to actually finish the commit
362
What should be done with the text of modified files while this is
363
underway? I don't think we want to count on holding them in memory
364
and we can't trust the working files to stay in one place so I
365
suppose we need to move them into the text store, or otherwise into
366
a temporary directory.
368
If the commit does not actually complete, we would rather the
369
content was not left behind in the stores.
221
373
* GUI (maybe in Python GTK+?)
added
removed
TODO
