« back to all changes in this revision
Viewing changes to TODO
- browse files at revision 2382.2.5
- compare with another revision
- download diff
- download tarball
- view history from revision 2382.2.5
- Committer: Martin Pool
- Date: 2007-04-04 06:17:31 UTC
- mto: This revision was merged to the branch mainline in revision 2397.
- Revision ID: mbp@sourcefrog.net-20070404061731-tt2xrzllqhbodn83
Contents of TODO file moved into bug tracker
- files modified:
- TODO
added
removed
TODO
1
* quotefn doubles all backslashes on Windows; this is probably not the
2
best thing to do. What would be a better way to safely represent
3
filenames? Perhaps we could doublequote things containing spaces,
4
on the principle that filenames containing quotes are unlikely?
5
Nice for humans; less good for machine parsing.
6
7
* Branch.update_revisions() inefficiently fetches revisions from the
8
remote server twice; once to find out what text and inventory they
9
need and then again to actually get the thing. This is a bit
10
inefficient.
11
12
One complicating factor here is that we don't really want to have
13
revisions present in the revision-store until all their constituent
14
parts are also stored.
15
16
The basic problem is that RemoteBranch.get_revision() and similar
17
methods return object, but what we really want is the raw XML, which
18
can be popped into our own store. That needs to be refactored.
19
20
* Guard against repeatedly merging any particular patch.
21
22
* More options for diff:
23
24
- diff two revisions of the same tree
25
26
- diff two different branches, optionally at different revisions
27
28
- diff a particular file in another tree against the corresponding
29
version in this tree (which should be the default if the second
30
parameter is a tree root)
31
32
- diff everything under a particular directory, in any of the above
33
ways
34
35
- diff two files inside the same tree, even if they have different
36
ids
37
38
- and, of course, tests for all this
39
40
* Reproducible performance benchmark to measure whether performance is
41
getting better or worse.
42
43
* ``bzr log -m foo`` should perhaps error if nothing matches?
44
45
* ``bzr diff -r 30 -r 40 foo.c`` or ``bzr diff -r30..40 foo.c``
46
47
If diffing between two branches then we probably want two -r
48
options, since the revisions don't form a range that can be
49
evaluated on either one.
50
51
* bzr diff shouldn't diff binary files
52
53
* setup.py install when run from a bzr tree should freeze the tree
54
revision-id into the installed bzr.
55
56
* bzr script should trap ImportError and perhaps give a better error
57
message?
58
59
* revert after a merge should possibly remove all the BASE/THIS/OTHER
60
files to get you back to where you were.
61
62
* files that are added and then deleted are still reported as added
63
64
* stores should raise KeyError, not IndexError
65
66
* merging from a remote branch seems to sometimes raise errors not
67
present locally
68
69
* should be possible to give a related branch when pulling from a
70
remote branch to make things faster
71
72
* sometimes gives "conflicting add" even when the contents are in fact
73
the same???
74
75
* BZRDIR should be in branch.py not __init__.py.
76
77
* bzrlib.delta.compare_trees() shouldn't need to sort all of the result
78
lists, since they are being added in alphabetical order. Write tests
79
to guarantee this, and remove the sort() calls.
80
81
Medium things
82
-------------
83
84
* merge should add all revision and inventory XML to the local store.
85
86
* check should give a warning for revisions that are named in the
87
chain but not actually present in the store.
88
89
* remove anything outside of the branch implementation that directly
90
accesses the stores.
91
92
* More efficient diff of only selected files. We should be able to
93
just get the id for the selected files, look up their location and
94
diff just those files. No need to traverse the entire inventories.
95
96
* Fix up Inventory objects to represent root object as an entry.
97
98
* Don't convert entire entry from ElementTree to an object when it is
99
read in, but rather wait until the program actually wants to know
100
about that node.
101
102
* Extract changes from one revision to the next to a text form
103
suitable for transmission over email.
104
105
* More test cases.
106
107
- ``missing`` command
108
109
- Selected-file commit
110
111
- Impossible selected-file commit: adding things in non-versioned
112
directories, crossing renames, etc.
113
114
* Write a reproducible benchmark, perhaps importing various kernel versions.
115
116
* Directly import diffs! It seems a bit redundant to need to rescan
117
the directory to work out what files diff added/deleted/changed when
118
all the information is there in the diff in the first place.
119
Getting the exact behaviour for added/deleted subdirectories etc
120
might be hard.
121
122
At the very least we could run diffstat over the diff, or perhaps
123
read the status output from patch. Just knowing which files might
124
be modified would be enough to guide the add and commit.
125
126
Given this we might be able to import patches at 1/second or better.
127
128
* revfile compression.
129
130
* Split inventory into per-directory files?
131
132
* Fix ignore file parsing:
133
134
- fnmatch is not the same as unix patterns
135
136
- perhaps add extended globs from rsh/rsync
137
138
- perhaps a pattern that matches only directories or non-directories
139
140
* Consider using Python logging library as well as/instead of
141
bzrlib.trace.
142
143
* Commands should give some progress indication by default.
144
145
- But quieten this with ``--silent``.
146
147
* Change to using gettext message localization.
148
149
* Make a clearer separation between internal and external bzrlib
150
interfaces. Make internal interfaces use protected names. Write at
151
least some documentation for those APIs, probably as docstrings.
152
153
Consider using ZopeInterface definitions for the external interface;
154
I think these are already used in PyBaz. They allow automatic
155
checking of the interface but may be unfamiliar to general Python
156
developers, so I'm not really keen.
157
158
* Commands to dump out all command help into a manpage or HTML file or
159
whatever.
160
161
* Handle symlinks in the working directory; at the very least it
162
should be possible for them to be present and ignored/unknown
163
without causing assertion failures.
164
165
Eventually symlinks should be versioned.
166
167
* Allow init in a subdirectory to create a nested repository, but only
168
if the subdirectory is not already versioned. Perhaps also require
169
a ``--nested`` to protect against confusion.
170
171
* Branch names?
172
173
* More test framework:
174
175
- Class that describes the state of a working tree so we can just
176
assert it's equal.
177
178
* Try using XSLT to add some formatting to REST-generated HTML. Or
179
maybe write a small Python program that specifies a header and foot
180
for the pages and calls into the docutils libraries.
181
182
* --format=xml for log, status and other commands.
183
184
* Attempting to explicitly add a file that's already added should give
185
a warning; however there should be no warning for directories (since
186
we scan for new children) or files encountered in a directory that's
187
being scanned.
188
189
* Better handling of possible collisions on case-losing filesystems;
190
make sure a single file does not get added twice under different
191
names.
192
193
* Clean up XML inventory:
194
195
- Use nesting rather than parent_id pointers.
196
197
- Hold the ElementTree in memory in the Inventory object and work
198
directly on that, rather than converting into Python objects every
199
time it is read in. Probably still expose it through some kind of
200
object interface though, but perhaps that should just be a proxy
201
for the elements.
202
203
- Less special cases for the root directory.
204
205
* Perhaps inventories should remember the revision in which each file
206
was last changed, as well as its current state? This is a bit
207
redundant but might often be interested to know.
208
209
* stat cache should perhaps only stat files as necessary, rather than
210
doing them all up-front. On the other hand, that disallows the
211
optimization of stating them in inode order.
212
213
* It'd be nice to pipeline multiple HTTP requests. Often we can
214
predict what will be wanted in future: all revisions, or all texts
215
in a particular revision, etc.
216
217
urlgrabber's docs say they are working on batched downloads; we
218
could perhaps ride on that or just create a background thread (ew).
219
220
* Paranoid mode where we never trust SHA-1 matches.
221
222
* --dry-run mode for commit? (Or maybe just run with
223
check-command=false?)
224
225
* Generally, be a bit more verbose unless --silent is specified.
226
227
* Function that finds all changes to files under a given directory;
228
perhaps log should use this if a directory is given.
229
230
* XML attributes might have trouble with filenames containing \n and
231
\r. Do we really want to support this? I think perhaps not.
232
233
* Unify smart_add and plain Branch.add(); perhaps smart_add should
234
just build a list of files to add and pass that to the regular add
235
function.
236
237
* Function to list a directory, saying in which revision each file was
238
last modified. Useful for web and GUI interfaces, and slow to
239
compute one file at a time.
240
241
This will be done when we track file texts by referring to the
242
version that created them.
243
244
* Check locking is correct during merge-related operations.
245
246
* Perhaps attempts to get locks should timeout after some period of
247
time, or at least display a progress message.
248
249
* Don't pass around command classes but rather pass objects. This'd
250
make it cleaner to construct objects wrapping external commands.
251
252
* Track all merged-in revisions in a versioned add-only metafile.
253
254
* ``uncommit`` command that removes a revision from the end of the
255
revision-history; just doing this is enough to remove the commit,
256
and a new commit will automatically be made against the
257
predecessor. This can be repeated.
258
259
It only makes sense to delete from the tail of history.
260
261
This has been implemented, but it does not remove the texts from
262
the store.
263
264
* ``bzrlib.lazy_import.lazy_import`` could check to see if a given
265
module has already been imported (look at ``sys.modules``). If it
266
has, then just import it directly, rather than creating a lazy
267
object.
268
269
270
Large things
271
------------
272
273
* Generate annotations from current file relative to previous
274
annotations.
275
276
- Is it necessary to store any kind of annotation where data was
277
deleted?
278
279
* Update revfile_ format and make it active:
280
281
- Texts should be identified by something keyed on the revision, not
282
an individual text-id. This is much more useful for annotate I
283
think; we want to map back to the revision that last changed it.
284
285
- Access revfile revisions through the Tree/Store classes.
286
287
- Check them from check commands.
288
289
- Store annotations.
290
291
.. _revfile: revfile.html
292
293
* Hooks for pre-commit, post-commit, etc.
294
295
Consider the security implications; probably should not enable hooks
296
for remotely-fetched branches by default.
297
298
* Pre-commit check. If this hook is defined, it needs to be handled
299
specially: create a temporary directory containing the tree as it
300
will be after the commit. This means excluding any ignored/unknown
301
files, and respecting selective commits. Run the pre-commit check
302
(e.g. compile and run test suite) in there.
303
304
Possibly this should be done by splitting the commit function into
305
several parts (under a single interface). It is already rather
306
large. Decomposition:
307
308
- find tree modifications and prepare in-memory inventory
309
310
- export that inventory to a temporary directory
311
312
- run the test in that temporary directory
313
314
- if that succeeded, continue to actually finish the commit
315
316
What should be done with the text of modified files while this is
317
underway? I don't think we want to count on holding them in memory
318
and we can't trust the working files to stay in one place so I
319
suppose we need to move them into the text store, or otherwise into
320
a temporary directory.
321
322
If the commit does not actually complete, we would rather the
323
content was not left behind in the stores.
324
325
* Web interface
326
327
* GUI (maybe in Python GTK+?)
328
329
* C library interface
330
331
* Expansion of $Id$ keywords within working files. Perhaps do this in
332
exports first as a simpler case because then we don't need to deal
333
with removing the tags on the way back in.
334
335
* ``bzr find``
336
337
Selftest
338
--------
339
340
* Write some benchmark tests
341
342
* we could display the ids of some tests selected by time to rune, or cputime,
343
rusage.
344
345
* We could run tests multiple times and report deviation/slowest/fastest.
346
347
* Gather lsprofile output for each test individually - log that somewhere.
348
349
* Should plugins be able to offer benchmark tests ?
350
351
CHANGESETS
352
----------
353
Ensure changeset command behaves properly w/revisions (e.g. 0)
354
Figure out whether testament-related bug is my problem
355
Get rid of the 'old' directory
356
Bundle signatures
357
What to do about ghost ancestors in remote branch that are present locally?
1
For things to do in Bazaar development, see
2
3
https://bugs.launchpad.net/bzr/
4
https://blueprints.launchpad.net/bzr/
Loggerhead is a web-based interface for Breezy
Version: 2.0.1