redo 0.07: once upon a time
Avery Pennarun
officially name the version everybody has been running anyway for the
past several months. Since that includes 40 commits on top of the
official 0.06 release, I think it deserves a number. This version is
really just a bunch of bugfixes on top of 0.06 and so anyone who
really is still using 0.06 should find it harmless.
However, I encourage you to not use 0.07 since I'm about to send out
an announcement about 0.10 :)
Henry Gebhardt (1):
version/vars.do: avoid infinite loop if the tag string is " master"
Joseph Garvin (2):
Use /usr/bin/env to detect python location
Fix tests when the sqlite3 command line utility isn't installed.
Avery Pennarun (37):
vars.do: an improvement on the previous patch.
redo-ifchange: remove special case for zero arguments.
Simplify autodependency examples.
README: a bunch of improvements to answer recent questions.
README.md: typo fix for my previous fix.
README.md: mention the mailing list nearer to the top.
Rename flush-cache.py to flush-cache, and rearrange the code a bit.
minimal/do: 'redo-ifchange' with no parameters shouldn't try to build all.
redo-sh: on Solaris, try /usr/xpg4/bin/sh for a POSIX shell.
minimal/do: work around a bug with empty .do files.
shelltest: detect the ". of empty script" shell problem.
shelltest.od: test that $1/$2/$3 are set correctly in "." scripts.
redo-sh.do: add 'posh' as another possible shell.
MacOS: "-e filename/." returns true even for non-directories.
minimal/do: leak fewer variables, especially PREFIX.
Test for previous PREFIX fix to minimal/do.
minimal/do: fix a typo: DO_BUILD -> DO_BUILT
t/curse/Makefile: delete.
builder.py: detect overrides by only ctime, not all of struct stat.
Test for previous atime/ctime fix.
md2man: fix compatibility problem with python-markdown 1.6.
redo: only default to 'all' in the toplevel instance of redo.
Don't crash on targets in directories that don't exist yet.
Add a test for the preceding fix (nonexistent directories).
state.relpath: correct broken paths ending with /, etc.
Test for precenting state.relpath fix.
User-overridden targets stay overridden even if the last build failed.
Test for previous override fix.
redo-sh.do: prettier log messages.
shelltest.od: a few more/better tests and clarification comments.
t/shelltest.od: another test for ${x#y} behaviour.
shelltest: use a more magical magic number for success.
Don't use "insert ... default values" in sqlite3.
t/shelltest.od: quiet some error messages that get printed during testing.
redo-sh: reverse a case that caused 'set -e' to abort on zsh.
Docs: recommend adding redo-ifchange with a lot of targets at a time.
t/pytest.do: don't run in minimal/do.
Have fun,
Avery