summaryrefslogtreecommitdiff
path: root/lib/Parse
Commit message (Collapse)AuthorAge
* Release 0.47Kenichi Ishigaki2024-04-21
|
* Drop role if Feature::Compat::Class is used (for now)Kenichi Ishigaki2024-04-21
|
* Improve class detectionKenichi Ishigaki2024-04-21
|
* Release 0.46Kenichi Ishigaki2024-04-15
|
* Strip BOM before finding a packageKenichi Ishigaki2024-04-15
|
* Release 0.45Kenichi Ishigaki2024-01-21
|
* Make sure to stringify versionKenichi Ishigaki2024-01-21
|
* Release 0.44Kenichi Ishigaki2023-04-27
|
* Support class NAME VERSION and role NAME VERSION (for the latest ↵Kenichi Ishigaki2023-04-27
| | | | perl/Object::Pad)
* Release 0.43Kenichi Ishigaki2020-12-12
|
* Avoid instantiating `Safe` compartment if operating in "unsafe" mode.Graham TerMarsch2020-12-10
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When operating in "unsafe" mode, the `$comp` never ends up getting used. So, why even instantiate/initialize it if we know that we're not going to be needing it? This patch results from an examination of the time spent performing an install of `Paws` (via `cpm`), and finding that a substantial amount of overhead was involved in simply instantiating/initializing the `Safe` compartment, even when `Parse::PMFile` was being used in `UNSAFE => 1` mode. While for many distributions, the overhead involved here would be negligible, for `Paws` (with over 20k `.pm` files), the overhead becomes substantial. Rough timings with `cpm install --verbose Paws`... Upstream: 16041 DONE resolve (0.477sec) Paws -> Paws-0.42 (from MetaDB) 16041 DONE fetch (19.435sec) Paws-0.42 16041 DONE configure (25.846sec) Paws-0.42 16041 DONE install (40.388sec) Paws-0.42 With Patch: 72651 DONE resolve (0.455sec) Paws -> Paws-0.42 (from MetaDB) 72651 DONE fetch (6.140sec) Paws-0.42 72651 DONE configure (4.488sec) Paws-0.42 72651 DONE install (41.798sec) Paws-0.42
* release 0.42Kenichi Ishigaki2019-11-09
|
* releng 0.41Kenichi Ishigaki2016-11-03
|
* not to count package declaration in a stringKenichi Ishigaki2016-11-03
|
* Add global modifier to package name substitutionRob Halliday2016-11-03
| | | | | | | | In _packages_per_pmfile package names get normalised to the arisdottle format so a package My'Test would be normalised to My::Test. Unfortunately My'Test'Package would get normalised to My::Test'Package rather than My::Test::Package. Add the global modifier to the package name substitution fixes this.
* omit warning categories as older versions of version does not register a ↵Kenichi Ishigaki2016-02-21
| | | | version category
* releng 0.40Kenichi Ishigaki2016-02-21
|
* silenced 'alpha->numify() is lossy' warningKenichi Ishigaki2016-02-21
|
* releng 0.39Kenichi Ishigaki2016-01-10
|
* releng 0.38Kenichi Ishigaki2016-01-09
|
* releng 0.37Kenichi Ishigaki2016-01-07
|
* ported version_ok() implemented in PAUSEKenichi Ishigaki2016-01-07
|
* releng 0.36Kenichi Ishigaki2015-04-16
|
* remove an unnecessary $DB::single line (remnant of PAUSE's debugging, ↵Kenichi Ishigaki2015-04-16
| | | | spotted by oalders++)
* releng 0.35Kenichi Ishigaki2015-01-17
|
* tolerate version line errors as undef (fix #8)Kenichi Ishigaki2015-01-17
|
* releng 0.34Kenichi Ishigaki2015-01-04
|
* use __clean_eval() as Module::Metadata doesKenichi Ishigaki2015-01-04
| | | | | - to localize $VERSION properly when evaluating version line - https://github.com/Perl-Toolchain-Gang/Module-Metadata/commit/2e9ed0c20f09085b4884c9140569664d38eccdb3
* releng 0.33Kenichi Ishigaki2014-12-13
|
* Revert "restore what are actually overloaded", which seems to have caused ↵Kenichi Ishigaki2014-12-13
| | | | | | | | | segfaults under some environments. This reverts commit ebba1f6486ce1662e55c23a311ed73140f809d13. Conflicts: lib/Parse/PMFile.pm
* releng 0.32Kenichi Ishigaki2014-12-12
|
* not to redo overload->import for version (broken under 5.12)Kenichi Ishigaki2014-12-12
|
* restore what are actually overloadedKenichi Ishigaki2014-12-12
|
* numify earlier for Safe not to handle a version object in revalKenichi Ishigaki2014-12-12
|
* releng 0.31Kenichi Ishigaki2014-12-09
|
* added tests for version->declare/qv without using version explicitlyKenichi Ishigaki2014-12-09
|
* releng 0.30Kenichi Ishigaki2014-12-05
|
* reimport overload into version::vpp only if 'use version' statement doesn't ↵Kenichi Ishigaki2014-12-05
| | | | exist in the VERSION line
* Revert "not to import overload into version::vpp explicitly (hopefully for ↵Kenichi Ishigaki2014-12-04
| | | | | | perl 5.12-16)" This reverts commit c55b33dfeb3ca719719c1cde5dd49a19a053b4d3.
* not to import overload into version::vpp explicitly (hopefully for perl 5.12-16)Kenichi Ishigaki2014-12-04
|
* restore overloaded stuff of all the loaded version modulesKenichi Ishigaki2014-12-04
| | | | | - instead of the one used as version(.pm) - RT-100696: Sharing *version:: into Safe compartment breaks version::vpp overloads (haarg++)
* releng 0.29Kenichi Ishigaki2014-10-10
|
* use default namespace to suppress a warning on UNIVERSAL::VERSION redefinitionKenichi Ishigaki2014-10-10
| | | | - Safe.pm has an internal 'Erase' flag and doesn't erase stuff it added if an explicit namespace is passed to new()
* releng 0.28Kenichi Ishigaki2014-10-08
|
* let's see...Kenichi Ishigaki2014-10-08
|
* import qv if UNSAFEKenichi Ishigaki2014-10-08
|
* releng 0.27Kenichi Ishigaki2014-10-08
|
* added UNSAFE option as the last resort for older perlsKenichi Ishigaki2014-10-08
|
* replaced $fh->close with close $fh for older perlsKenichi Ishigaki2014-10-07
|
* removed unused modulesKenichi Ishigaki2014-10-07
|
generated by cgit v1.2.3 (git 2.25.1) at 2024-06-16 13:08:49 +0000