On win32, the command line arguments seems to be limited about 2400
bytes. So 'add', 'delete', 'propget' and 'propdel' operations are
called several times within this limitation. And also 'commit'
operation is changed to be called without passing target files not to
exceeds this limitation because it cannot be a separated like other
operations above.
These options make qnew add and qrefresh update a "# Date "-style
header line. This allows proper recording of creation / last
modification dates of patches in patch queues.
Note that `qrefresh -D` only updates existing header lines. It never
adds them, and does not warn about this. This is because I expect
people to have `[default] qrefresh -D` in their .hgrc so patches with
tracked dates get updated, others are left unchanged. The suggested
setup in .hgrc is, in fact,
[default]
qnew = -D -U
qrefresh = -D
I tried to not mix header styles, so `qnew -D -U` now writes the user in
"# User "-style, while `qnew -U` still writes it "From: "-style. Also, if
`qrefresh -U` must add the user, it does so in "# User "-style if the
header contains a "# HG changeset patch" line. (This is caused by mq
not supporting the "Date: "-style header line at all - a reasonable choice
given its standard date format.)
63beab327d26 introduced using ui.username before web.contact, but this was
never documented and might cause commit accidents.
- Drop web.author (deprecated since 2005)
- Try ui.username or $EMAIL as a fallback to display something useful.
- Update docs for the fallbacks.
Automatically detect whether we're looking for a bad to good
transition rather than the usual good to bad transition by detecting
when badrev is inside the good set and flipping good/bad.
Once an ancestor list has grown past the perfect threshold, all
descendants are less optimal. Use a poison dict to avoid pointless
operations on their long ancestor lists, thus eliminating most of the
work.
- calculate the children of all candidates
- for each candidate, combine ancestor lists
- pass ancestor lists to children
- store ancestor count
This eliminates the O(n**2) memory usage, while maintaining about the
same performance.
- use "branch 'foo'" to distinguish from "branch merge".
- commit messags can be empty (to abort commits)
- Added value for editor message: Tell about HG: lines like CVS does.
Python already lets one to embed RE flags directly in a regex, which
is a much nicer way to do things:
(?iLmsux)
(One or more letters from the set "i", "L", "m", "s", "u", "x".)
...
matchflags was introduced in 14a7e862c60f, and the record extension is the only
user. I've killed matchflag, and adjusted record code appropriately.
We can't store properties for files we haven't added to repo. Similarly,
we can't copy file to directory we haven't added to svn yet. Remember
needed changes and apply them in putcommit().