[Bps-public-commit] Config-GitLike annotated tag, 1.00, created. 1.00
spang at bestpractical.com
spang at bestpractical.com
Tue Aug 11 12:20:52 EDT 2009
The annotated tag, 1.00 has been created
at 3c761e141de2000bb7d2f5ed14ba86065102c857 (tag)
tagging b31fef6931449da0f5a310eda497caa4d2fbe8de (commit)
tagged by Christine Spang
on Wed Jul 8 22:37:14 2009 +0200
- Log -----------------------------------------------------------------
Tagging version '1.00' using shipit.
Alex Vandiver (15):
Initial import
Allow for comments not after definitions
Add support for escaping characters when not in quotes
Fix case on stored values
Move to a callback-based parser, for ->set
Add offset calculation, so ->set can know where things are
Minor doc fixups
Preliminary (non-multiple) set support
Make ->load_dirs and ->load take which directory, defaulting to the cwd during ->load
Make Config::GitLike->load_dirs correctly walk up the directory tree
In ::Cascading, deeper dirs are loaded later and override earlier and shallower ones
Fix up a couple minor comment nits
For paramhash functions, explicitly list parameters and defaults
Make error callback get passed the original string and offset of error, not just remaining string
Ignore blib/ directory
Christine Spang (92):
typo fix in Makefile.PL
fix a syntax error
some pod while I'm figuring out how this stuff works anyway
a billion regexes without comments make me want to cry
stick an __END__ where the code ends and only doc remains
pod Config::GitLike::Cascaded as well
more doc love
more pod
fix for discrepency caught by git test suite: section names are case-insensitive
check to make sure key is given in set
fix missing , in format_definition which causes it to lose all its args
occurrance -> occurrence
attempt to load only plain files, not directories
fix for test suite compliance: reject keys that begin with a number
initial port of git config's test suite
add build deps to Makefile.PL
make cast's behaviour in-line with what the test suite specifies
this exception can also be thrown if there's a newline in the key
commit the script I used to translate some bits of the git config test suite for posterity
pass in tmpdir to the TestConfig constructors on the last two tests
parser fix: don't tack trailing whitespace onto values, even if the value is followed by a comment
make dump return a string rather than printing. this is more useful for things like tests.
un-TODO testing dump. since our interface doesn't keep track of key/value data, make the tests expect the output to be sorted
kill an unnecessary \
add .gitignore
pod updates
small comment tweaks in test suite
un-TODO rename_section tests to prep for implementation
pull config file reading out to _read_config
fix some test bits that the translation script (and/or my manual frobbing) b0rked
un-TODO tests for remove_section
add rename_section and remove_section and related changes
make this not quite so horrible to read
bugfix: compare variable names case-insensitively (and un-TODO the related test)
casting on set
kill commented-out NUL-byte termination tests; we're not going to implement that feature
int casting on set
bool-or-int
get_regexp (didn't implement this in ->get after all since it needs to return something completely different)
filter for set
multiple unset
multivar support
test that stuff like '[section] [section2] a = 1' parses correctly
I think I count as an author at this point
additional tests for casting/filtering with get_regexp and get_all
split up some long lines
smarter return values based on context for methods that return non-scalars and ->dump now prints rather than returns in void context
some small test cleanups
perltidy t1300-repo-config.t
use tests
kill t/00-basic.t since it doesn't do anything useful anymore
I'm beginning to agree with PBP in that non-contiguous POD = contorted code, confused docs, or both. Oh hey, and I overhauled most of the doc while I was at it.
saw something in the git config doc that I thought we should test
add version number to Makefile.PL
add some more files to .gitignore
fix broken pod
not cuddling elsif/else seems to make this look a bit cleaner
comment about killing leading whitespace on section removes
BUGS section in pod
test int/num casting
minor doc fixes
point out in the synopsis that you need to ->load before a ->get after a ->set
document ways we differ from git-config
Merge branch 'master' of fsck.com:/git/config-gitlike
keep track of what files have been loaded
fixes to load / load_dirs doc
keys can only contain alphanumeric and - characters (no underscores)
Dramatically relax rules for key names.
Make dump print multiple values correctly.
Make this legible.
Ow: clear multiple on reload as well.
Keep track of which file each key was loaded from.
Add tests for overriding keys from files of higher precedence.
Move Config::GitLike::Cascaded doc to the end of its file.
Don't call these small internal subs as methods.
Make set support setting a group of variables at once.
Terminology cleanup + make sure to compare subsections
User doc for group_set.
Allow enforcement of "only read/write things that git can" with
New error_callback function for parse_content.
A bugfix and regression tests for permissive parsing
Minor typo fix.
Die if we try to open a file for writing and fail.
Mention Prophet/SD in SEE ALSO
skeleton .shipit
Add inc/.author to .gitignore
Don't try to do clever things with locking as it seems to backfire.
Make _read_config and _write_config methods rather than functions so they're overridable
I don't actually want to announce to Twitter, I don't think
Right version number in Changes
Don't add Makefile.old to MANIFEST
don't commit MANIFEST(.bak)
Jesse Vincent (12):
Moved the test file to a test directory
Added a basic Makefile.PL
Snapshot the inc that goes with the Makefile.PL
Added a use 5.8 to make Module::Install happy
quiet down some diags
Added author tests
added authortests inc file
Moved the version number to the library
File::HomeDir isn't really acceptable software on OSX
Update inc::Module::Install
added Changes file
Manifest.SKIP updates
-----------------------------------------------------------------------
More information about the Bps-public-commit
mailing list