Annotation of wikisrc/users/dholland/mercurial.mdwn, revision 1.2
1.1 dholland 1: ## Using Mercurial and mq to work on NetBSD
2:
3: This page contains directions for using Mercurial as a commit buffer
4: for NetBSD.
5:
6: (It will not do you much good if you're trying to convert the master
7: NetBSD tree to Mercurial or to work with such a converted tree.)
8:
9: ### What it is
10:
11: Mercurial is a distributed version control system ("DVCS").
12: mq is an extension to Mercurial for handling patch queues.
13: The concept of patch queues was introduced by Quilt some years back.
14:
15: This document assumes you already know more or less how to use
16: Mercurial but may not have used mq before.
17:
18: ### The model we're using here
19:
20: What we're going to do is commit a NetBSD CVS working tree into a
21: Mercurial repository.
22: You can then use Mercurial to merge; it is better at this than CVS.
23: You can also commit changes locally and ship them back to the CVS
24: master later; this is useful in a variety of ways.
25: You can potentially also clone the Mercurial tree and work jointly
26: with other people, but there are limits to this as we'll discuss in a
27: moment.
28:
29: Because the NetBSD tree is rather large, you will find that if you
30: commit the whole thing into Mercurial that a number of operations
31: (anything that scans the working tree for changes) become annoyingly
32: slow.
33: It isn't slow enough to be unusable, and it's quite a bit faster than a
34: comparable CVS option (like running cvs update from the top level),
35: but it's slow enough to be annoying.
36:
37: For this reason, in most cases, I recommend committing only part of
38: the tree into Mercurial and telling it to ignore the rest. This means
39: you can't in general usefully clone the resulting Mercurial repo
40: (although that depends on exactly what you leave out) but this is not
41: a major problem unless you're specifically trying to work with someone
42: else.
43:
44: So the basic model here is that you check out a CVS working tree and
45: use Mercurial to manage local changes to part of it, then later on
46: commit those changes back to the master CVS repository.
47:
48: ### Branches vs. patches
49:
50: There are two ways you can manage your changes: as a branch, or as a
51: patch queue.
52: The advantage of a patch queue is that you can easily commit each
53: patch individually back to CVS, and you can go back and forth between
54: them and debug and polish each one separately.
55: The disadvantage is that the merge facilities are (as far as I know
56: anyway) relatively limited.
57:
58: Conversely, if you commit your changes to a branch, you get all the
59: native merging support in Mercurial.
60: However, it is painful to try to commit anything other than one big
61: diff for the whole branch back to CVS.
62: (You might be able to do it via bookmarks and rebasing, but I've never
63: tried and have no desire to figure out how.)
64:
65: If you don't want to keep the incremental history of your local
66: commits, use a branch.
67: If you do, use a patch queue.
68:
69: It is possible to use multiple branches to allow you to commit back in
70: several stages.
71: However, managing this is a major pain and I don't recommend it -- you
72: might get away with two branches but more than that is probably a bad
73: idea.
74:
75: There's a Mercurial extension called the "patch branch extension" that
76: lets you manage a whole graph of patches using branches.
77: I haven't tried using it in some years; at the time it had scaling
78: problems such that it became horrifyingly slow once you had more than
79: a handful of such branches.
80: That might have been improved in the meantime; if you find yourself
81: wanting to use both branches and patches, it might be worth looking
82: into.
83:
84: It is also fairly probable that there is now a solution for merging
85: with patch queues; it's been a while since I had time to look closely.
86:
87: ### Setting up
88:
89: First, check out a CVS working tree.
90: You probably want to use a different one for each project, because
91: different projects require changing different parts of the tree and so
92: you will probably want to have Mercurial ignore different subtrees for
93: different projects.
94: (At least, I find it so; it depends on what you're working on.)
95:
1.2 ! dholland 96: % cvs -d cvs.netbsd.org:/cvsroot checkout -P src
1.1 dholland 97:
98: Now create a Mercurial repository at the top level.
99: (If you are working only in a subtree and you are *sure* that you will
100: never need to change anything in other parts of the tree, you can
101: create the Mercurial repository in a subtree.
102: But unless you're absolutely certain, don't take the risk.)
103:
104: % cd src
105: % hg init
106:
107: If you're going to be using a patch queue, now enable mq.
108:
109: % vi .hg/hgrc
110: and add
111: [extensions]
112: hgext.mq =
113: (Since the extension is built into Mercurial, that's all you need.)
114: You can if you prefer also put this in your .hgrc so mq is always on.
115: Then do
116: hg qinit -c
117: The -c option tells mq that you'll be checkpointing your patches,
118: which is usually a good idea.
119:
120: Now prepare a .hgignore file.
121: This file contains one regular expression per line; Mercurial ignores
122: files (and subdirectories) whose paths from the repository root match
123: one of the regexps.
124: Add at least:
125: ^CVS$
126: /CVS$
127: to ignore all the CVS control directories in the CVS checkout.
128: While you can commit these to Mercurial, there's no point and it gets
129: awkward if owing to mistakes later you end up having to merge them.
130:
131: If you aren't arranging to put the tree's object directories somewhere
132: else, then also add
133: ^obj\.[0-9a-z]$
134: /obj\.[0-9a-z]$
135: and you might want
136: ^sys/arch/[0-9a-z]*/compile/[A-Z]
137: to ignore kernel build directories.
138:
139: Ignore subtrees that you aren't working in.
140: You don't have to bother to be very selective; the goal is to rapidly
141: rule out a few large subtrees that you definitely don't care about, in
142: order to avoid wasting time scanning them for changes.
143: Unless you plan to be working with 3rd-party software,
144:
145: ^external$
146: ^gnu/dist$
147:
148: is a good starting point.
149: Alternatively, if you aren't going to be working on MD kernel stuff or
150: bootloaders,
151:
152: ^sys/arch$
153:
154: is a good choice as it's also large.
155:
156: You can always unignore stuff later, so don't worry about remote
157: possibilities.
158:
159: Now commit the .hgignore file:
160:
161: % hg add .hgignore
162: % hg commit -m 'add .hgignore file' .hgignore
163:
164: Now add and commit the contents of the working tree:
165:
166: % hg add
167: % hg commit -m 'HEAD of 20130101'
168: (or whatever date)
169:
170: You are now in business.
171:
172: ### Working
173:
174: If you're using a branch, remember to change branches before you
175: commit anything:
176: % hg branch mystuff
177: You want to keep the default branch an untouched CVS tree so you can
178: use Mercurial to merge.
179: (And also so you can use Mercurial to extract diffs against CVS HEAD
180: and so forth.)
181:
182: Similarly, if you're using a patch queue, put everything in patches
183: and don't commit.
184: (There's a section below about working with mq if you aren't familiar
185: with it.)
186:
187: You can edit and build and test as normal.
188: Use hg commit or hg qrefresh to sync stuff into Mercurial.
189:
190: If you're using mq, it's a good idea to checkpoint your patch queue
191: periodically.
192: This is done as follows:
193: % hg qcommit
194: The patches directory (.hg/patches) is stored in its own Mercurial
195: repository, and this commits the patches to that repository.
196: If necessary you can then fetch older versions of the patches back and
197: so forth.
198:
199: ### Updating from CVS
200:
201: First, make sure all your changes are committed.
202: (If you have unfinished changes that aren't ready to commit, there's a
203: Mercurial extension for stashing them temporarily.
204: If you have stuff that you don't want to commit at all, like debugging
205: printouts or quick hacks, it's often convenient to keep those in their
206: own mq patch, even if you aren't using mq for development.)
207:
208: Now go back to a clean CVS tree.
209: If using branches, go back to the default branch:
210: % hg update -r default
211: If using mq, pop all the patches:
212: % hg qpop -a
213:
214: DO NOT run cvs update until/unless you have done this; it will make a
215: mess.
216: When you eventually do this by accident, see the section below on
217: recovering from mistakes.
218:
219: Now run cvs update from the top of the source tree:
220: % cvs -q update -dP
221:
222: You should get no conflicts from CVS and nothing should show as
223: modified.
224: (It is usually a good habit to save the cvs update output to a file to
225: be able to check this.)
226:
227: Tell hg to sync up:
228: % hg addremove
229:
230: Use hg to check what it thinks has changed:
231: % hg status
232:
233: Commit the changes to Mercurial:
234: % hg commit -m 'Updated to 20130202"
235:
236: Now you get to merge.
237:
238: If you're using a branch, you want to merge the changes into your
239: branch rather than merge your branch into the changes:
240: % hg update -r mystuff
241: % hg merge default
242: (edit and resolve as needed)
243: % hg commit -m 'sync with HEAD'
244:
245: If it tells you "update crosses branches" when trying to update back
246: to your branch, update to the parent changeset (the previous version
247: from CVS) first, as that's an ancestor of your branch.
248:
249: If you're using mq, the thing to do now is to push all your patches,
250: and if any reject, clean up the mess and refresh them.
251:
252: If patch tells you "hunk N succeeded at offset MMM with fuzz Q", it's
253: a good idea to manually inspect the results -- patch being what it is,
254: sometimes this means it's done the wrong thing.
255: Edit if needed.
256: Then (even if you didn't edit) refresh the patch so it won't happen
257: again.
258:
259: As I said above, it's quite likely that by now there's a better scheme
260: for merging with mq that I don't know about yet.
261:
262: ### Pushing back to CVS
263:
264: When you're ready to push your changes back to CVS (so they're really
265: committed), first (unless you're absolutely sure it's not necessary)
266: update from CVS as above and merge.
267: Then:
268:
269: If you're using a branch, go back to the default branch and merge your
270: changes into it:
271: % hg update -r default
272: % hg merge mystuff
273: % hg commit -m "prepare to commit back to cvs"
274: Now cvs add any new directories and files; be sure not to forget this.
275: It is a good idea to crosscheck with cvs diff and/or cvs update:
276: % cvs diff -up | less
277: % cvs -nq update -dP
278: Then you can cvs commit:
279: % cvs commit
280: Because of RCSIDs, committing into cvs changes the source files.
281: So now you need to do:
282: % hg commit -m 'cvs committed'
283: and if you intend to keep working in this tree, you want to merge that
284: changeset back into your branch to avoid having it cause merge
285: conflicts later.
286: Do that as above.
287:
288:
289: If you're using a patch queue, usually it's because you want to commit
290: each patch back to CVS individually.
291: First pop all the patches:
292: % hg qpop -a
293: Now, for each patch:
294: % hg qpush
295: % hg qfinish -a
296: % cvs commit
297: % hg commit -m "cvs committed previous"
298: With a long patch queue, you'll want to use the patch comments as the
299: CVS commit messages.
300: Also, running cvs commit from the top for every patch is horribly slow.
301: Both these problems can be fixed by putting the following in a script:
302: hg log -v -r. | sed '1,/^description:$/d' > patch-message
303: cat patch-message
304: echo -n 'cvs commit -F patch-message '
305: hg log -v -r. | grep '^files:' | sed 's/^files://'
306: (I call this "dogetpatch.sh") and then the procedure is:
307: % hg qpop -a
308: then for each patch:
309: % hg qpush && hg qfinish -a && dogetpatch.sh
310: % cvs commit [as directed]
311: % hg commit -m "cvs committed previous"
312: (This could be automated further but doing so seems unwise.)
313:
314: ### Using CVS within Mercurial
315:
316: You can successfully do any read-only CVS operation in the hybrid
317: tree: diff, annotate, log, update -p, etc.
318: Read-write operations should be avoided; if you mix upstream changes
319: with your changes you will find it much harder to commit upstream
320: later, and you may get weird merge conflicts or even accidentally
321: revert other people's changes and cause problems.
322:
323: If you clone the Mercurial tree and you didn't include the CVS control
324: files in it, you won't be able to do CVS operations from clones.
325: Including the CVS control files in the Mercurial tree is one way
326: around that.
327:
328: You will find that any large CVS operation on a clone is horribly
329: slow.
330: This is because making a clone causes CVS to think all the files in
331: the clone have been modified since you last ran it; it then re-fetches
332: every file you ask it about so it can update its own information.
333: For this reason cloning the Mercurial tree usually isn't worthwhile
334: and even when it is, including the CVS files in the Mercurial tree
335: isn't.
336:
337: Another consequence of this: do not try to cvs update in a cloned
338: Mercurial repository; use only the original.
339: Updating a clone basically downloads the entire tree over again from
340: the CVS server.
341:
342: DO NOT CVS COMMIT FROM A CLONE.
343: It is known that some operations that muck with the timestamps in a
344: CVS working tree can cause CVS to lose data.
345: It is not clear if hg clone is such an operation; don't be the person
346: who finds out the hard way.
347:
348: ### Recovering from mistakes
349:
350: The most common mistake is CVS updating when the Mercurial tree is not
351: in the proper state from that; e.g. onto your branch or while you have
352: patches applied.
353:
354: The basic strategy for this is to use hg revert to restore the part of
355: the tree it knows about, then go back to CVS, clean up the mess there,
356: and update properly.
357:
358: If you're using a branch:
359: % hg revert -C
360: % hg update -r default
361: If you're using a patch queue:
362: % hg revert -C
363: % hg qpop -a
364:
365: The problem is, CVS will now think you've changed every file that
366: Mercurial is managing, and the modifications are to revert all the
367: changes that have happened since your previous update.
368: You do *not* want that to turn into reality.
369: Hunt down (with cvs -n update) any files that CVS thinks are modified,
370: then rm them and run cvs update on them.
371: CVS will print "Warning: foo was lost" and restore an unmodified copy.
372:
373: When you have no files left that CVS thinks are modified, do a CVS
374: update on the whole tree and merge it as described above.
375: (You must do this, as the parts of the tree that Mercurial is ignoring
376: will otherwise be out of sync with the parts it's managing.)
377:
378: If you stored the CVS control files in Mercurial, then the revert will
379: restore them, but your tree will still be inconsistent so you still
380: need to do a proper update and merge immediately.
381:
382: ### mq
383:
384: The basic idea of mq (like quilt) is it maintains a series of patches
385: against the source tree, that are to be applied in order.
386: By applying them or removing them one at a time, you can move the tree
387: to any intermediate state; and then you can update the topmost patch,
388: insert a new patch, or whatever.
389:
390: To see the list of patches:
391: % hg qseries
392:
393: To apply the next patch:
394: % hg qpush
395:
396: To remove the current patch:
397: % hg qpop
398:
399: To merge current working tree changes into the current patch:
400: % hg qrefresh
401:
402: To also update the current patch's change comment:
403: % hg qrefresh -e
404:
405: To collect current working tree changes (if any) into a new patch:
406: % hg qnew PATCHNAME
407:
408: When there's an mq patch applied, you can't commit.
409: (Doing qrefresh is basically equivalent to committing the current
410: patch.)
411: Diff will show the changes against the last refreshed version of the
412: current patch; to see the complete changes for the current patch
413: (including current changes), use "hg qdiff".
414:
415: You can delete patches with "hg qrm" and rename them with "hg qmv".
416:
417: Patches are applied with patch, unfortunately, which means that if
418: they don't apply (which can happen if you or someone else changes
419: something under one) you get .rej files you have to clean up by hand
420: rather than a Mercurial merge.
421:
422: When a patch is ready to be committed for real, you do "hg qfinish" on
423: it.
424: This removes it from the patch queue and converts it to a normal
425: Mercurial changeset.
426:
427: To change the ordering of patches, you edit the file
428: .hg/patches/series.
429: If the patches aren't orthogonal you'll have to fix the rejections
430: when you next apply them.
431: (Don't do this with patches that are currently applied.)
432:
433: Use "hg help mq" to see the full list of mq-related commands.
434:
435: I'm sure there are better mq tutorials out there.
436:
437: ### Using mq
438:
439: The basic process when using mq is that you start a new patch, edit
440: and hack for a while, use hg qrefresh to commit it (once or many
441: times), and when you're done go on to the next one.
442:
443: If you find a bug in an earlier patch, you can go back to the patch
444: that introduced it and fix the bug there, creating a new version of
445: the offending patch that no longer contains the bug.
446: (Or you can create a new patch that fixes the bug, but insert it
447: immediately after the patch that created the bug.)
448:
449: When a patch is ready to be seen by other people, you "finish" it and
450: then it becomes a normal immutable changeset.
451:
452: One catch is that you can't push or pop the patch queue while you have
453: unsynced (uncommitted) changes.
454: There are two ways around this; there's a separate "stash" extension
455: that lets you put unfinished changes aside while you do something else.
456: Or, alternatively, you can create a new temporary patch holding your
457: unfinished changes, and then later use hg qfold to combine that with
458: the patch you originally meant this for.
459:
460: A variant of this problem is when you discover a bug, open an editor,
461: fix it, and then realize that you wanted to make the edit in an
462: earlier patch.
463: Then you go to pop the queue and it complains that you have a modified
464: file.
465: If the modification in question is the only uncommitted change, the
466: best way to deal with this is to create a new patch for it, then pop
467: to where you wanted it to go and use hg qfold to apply it there.
CVSweb for NetBSD wikisrc <wikimaster@NetBSD.org> software: FreeBSD-CVSweb