It has the unfortunate property that wrong configs seem to tend to
drop to the debugger while building, which hangs bulk builds and needs
to be masked. However, mucking with configurations isn't all that
difficult provided you can test them, and many that could be added
will probably just work.
Source Highlight is a library to format code written in a number
of programming languages as text in a number of markup languages.
This binding to the underlying C++ library is very basic, supporting
only the essential functionality.
Note that any exceptions thrown by the underlying C++ library are
caught by the Perl binding and rethrown using the equivalent of a
die statement.
Progress bars are handy - they tell you how much work has been done,
how much is left to do and estimate how long it will take.
But they can be fiddly!
This module does the right thing in almost all cases in a really convenient way.
platforms that can run i386-linux binaries; we don't need to set
ONLY_FOR_PLATFORM as well. (And especially not more restrictively than
necessary, such as excluding Linux.)
- Requires ExtUtils::ParseXS 3.18 because of 2.21 issues
(https://github.com/xslate/p5-Text-Xslate/issues/104)
- Support PUREPERL_ONLY=1
- Remove experimental requires_xs() because Module::Install
provides another requires_xs()
- Move t/01_example.t to xt/
of platforms accepted: the situation is that the upstream file is
binary and only semi-portable. It in particular won't work on netbsd-6
and up, regardless of machine type, because it uses time_t.
If anyone feels like improving the situation, get upstream to
distribute a portable file...
- Newlines should be consistently UNIX-stylee now in generated files (MITHALDU)
- Mark two tests as TODO if EUMM > 6.98 until utf8 gets fixed (BINGOS)
- Revert license behaviour (TSIBLEY)
- Documentation also mentioning metacpan.org (LLAP)
- Minimum perl version is now v5.6.0 (BOWTIE)
- Update license tables (BOWTIE)
- Prereqs and recommends adjusted (BOWTIE)
- Typo fixes (BOWTIE)
- Pod fixes (GREGOA)
- Fix make upgradedeps work with bundles (JNWHITLEY)
- Fix name_from regex to work with recent package syntax (BOWTIE)
- Performance enhancements when running Makefile.PL (GFUJI)
- Removed use of sub prototypes (BOWTIE)
- Remove LWP::UserAgent from the prereqs (TOKUHIROM)
- Require a more recent version of Module::ScanDeps (CHORNY)
- Fix failing test case with recent Parse::CPAN::Meta (BINGOS)
- scandeps.pl: die if an option is not recognized
- Reformat Changes file according to CPAN::Changes::Spec
- Modify %Preload rule: let Unicode::UCD explicitly imply utf8.pm.
This fixes PAR::Packer's self test.
Previously Unicode::UCD implied utf8.pm implicitly because
it contains calls to some utf8::foo() functions.
- Add %Preload rule: Mozilla::CA requires its cacert.pem file
- Recognize "do filename" constructs even if "do" isn't at the start
of a chunk.
- Upgrade to Module::Install 1.14
- Fix RT#98938: recognize Module::Runtime module-loading functions
- Fix a nasty typo that broke scandeps.pl option -E
$ scandeps -E "some string"
Unknown option: E
Can't open some string: No such file or directory at scandeps.pl line 49.
- also scandeps.pl: die if an option is not recognized
- Remove some overzealous heuristics from scan_chunk()
- they were looking for
Foo::Bar->something
Foo::Bar::whatever(...)
_anywhere_ in programs to infer a dependency on Foo/Bar.pm.
BEWARE: This might break some use cases, i.e. missing some dependencies.
On the other hand, this causes hard to investigate problems like the one
starting at http://www.mail-archive.com/par@perl.org/msg05531.html.
While the former can easily be worked around by the user itself (just
add a missing dependecy explicitly, e.g. using "pp -M ...") and
typically can be solved in general by adding a %Preload rule,
the latter just wastes people's times.
- Recognize Test::More require_ok() and use_ok()
- makes 3-static_oo_interface_real.t pass again (fallout from the above)
- Upgrade to Module::Install 1.12
- Add option -T to request information from CPAN
- don't access CPAN behind the user's back just because they have
CPANPLUS installed (it was in the Perl core from 5.10 to 5.18) -
it might not even have been configured (e.g. in a corporate internet)
- only do this when explicitly requested
- Fix RT #98203: Migrate from deprecated Module::Build::ModuleInfo to Module::Metadata
- thanx Petr Pisar (ppisar@redhat.com) for the hint
- add long option names to scandeps.pl
- implement option --xargs for scandeps.pl
- fix wrong version numbers in Changes
- Fix RT #92860 (t/7-check-dynaloader.t doesn't handle systems with mod2fname),
also RT #97519 (Fix for t/7-check-dynaloader.t on systems with DynaLoader::mod2fname)
- applied patch from Brian Fraser (fraserbn@gmail.com), thanks!
- lib/Module/ScanDeps/DataFeed.pm: apply here, too
- Fix recognition of (open() arguments) "<:encoding(klingon)",
implies modules PerlIO and PerlIO::encoding.
- Fix RT #90869: Use of uninitialized value $module in substitution (s///)
- Fix RT #87775: typo fixes, thanks dsteinbrunner@pobox.com
- new %Preload rule for B::Hooks::EndOfScope
- new %Preload rule for Pod::Usage
- add a fake %Preload rule that warns if use of Module::Implementation
or Module::Runtime is detected (coz' they're doing runtime loading)
- change some tests to use Test::Requires instead of homegrown stuff;
hence add it to "test_requires"
- clean up some uses of Test::More
- Fix RT #89000: test broken by indirect base.pm disuse
- delete base.pm from list of expected deps,
patch by Andrew Main (zefram@fysh.org)
- new %Preload rule for Net::HTTPS (e.g. used by LWP::Protocol::https)
- look for IO::Socket::SSL or Net::SSL
- new %Preload rule for YAML::Any
- try to figure out what YAML::Any would have used
(using YAML::Any->implementation)
- as fallback, include anything below YAML
- add %Preload rule for Params::Validate to detect
its PP and XS implementations
- Fix RT #80276 Module DateTime::Format::ISO8601 generates error
after being packaged
- caused by failing to pack DateTime::Format::Builder::Parser::XXX modules
needed by DateTime::Format::Builder::Parser
- add a corresponding %Preload rule
- update to Module::Install 1.06
(As of the last I remember looking at lcc, which was admittedly a
while ago, if it emits code for the wrong arch it's your fault, i.e. a
packaging/configuration problem.)
