Many thanks for the new release Autrijus! The -M fix solved the dependency issues I was having. Great job!
Clay. > -----Original Message----- > From: Autrijus Tang [mailto:[EMAIL PROTECTED] > Sent: Friday, February 27, 2004 11:18 AM > To: [EMAIL PROTECTED] > Subject: PAR 0.79_98 released! > > Here is PAR 0.79_98, the "testing waters" release, freshly > uploaded to CPAN: > > http://aut.dyndns.org/dist/PAR-0.79_98.tar.gz > MD5 (PAR-0.79_98.tar.gz) = 71c28cf213540e6d9b9d5444d4aa7f1a > > Pre-built Win32 binaries may be downloaded from CPAN and > http://aut.dyndns.org/par/. As usual, you still need to download and > unpack PAR-0.79_98.tar.gz, put the prebuilt .par file inside the extracted > PAR-0.79_98 directory, then proceed to "perl Makefile.PL". > > There are, some known issues that prevented this release from becoming > 0.80: > > - IO.dll version incompatibility is still unresolved. > - Modifying Tk icons on ActivePerl 5.8.3 fails here. > - "make all" may not install dependencies first. The workaround > is to "make config" first. > - Linux with GCC3 reportedly segfaults; a backtrace will be appreciated. > (To get a debug built, edit myldr/Makefile.PL to set the debug flag.) > - Rafael reports that 5.9.1 does not build cleanly, but it works fine > on FreeBSD here, so it may or may not be a Linux-specific problem. > > I expect a 0.79_99 with those issues resolved soon, and 0.80 shall follow > provided 0.79_99 does not introduce new regressions. Also, I'd like to > merge in Macolm's excellent regression test suite for PAR during 0.79_99. > > If Alan/Nicholas/Anybody manages to come up with recursive par_rmtmpdir() > implementation(s) for Win32/*nix, I'd be happy to merge them too. > > So... Please test this revision, so 0.80 will seriously rock. :-) > > [Changes for 0.80 - tentative] > > * New Features > > - Much better support for "pp -i" and "pp -N" (--icon/--info) > using the Win32::Exe module. You may now use EXE and DLL as > icon files. > > - If PAR_GLOBAL_CLEAN (-C, --clean) is not set, we now preemptively > extracts files under the cache directory. That made POSIX.pm > and other modules that depends on %INC pointing to real files > work correctly. > > - Now uses SHA-1 to create temporary directories and files, > instead of mtime. > > - Verbosity level is now 1..3, not 0..5; "pp -v" now takes > an optional integer, so "pp -v input.pl" is no longer an error. > > - New flags "-vv" and "-vvv", as shorthands for "-v 2" and "-v 3". > > - The user-settable PAR_CLEAN and PAR_TEMP environment variables has > been renamed to PAR_GLOBAL_CLEAN and PAR_GLOBAL_TEMP; the original > variables is still preserved within the program. This is so that a > pp-generated program can exec() or system() another one without > crippling its environment variables. > > - File lookups are now case-insensitive on case-insensitive > filesystems. > > * Bug fixes > > - "pp -d" (--dependent) prevented "pp -C" (--clean) from working. > > - The "pp -m" (--multiarch) option was implemented incorrectly > and thus broken. > > - Many documentation tweaks. > > - Previously, "pp -M" (--module) did not add the module itself, > only its dependencies. > > - Suppress a bogus warning when $ENV{$Config{ldlibpthname}} is empty. > > - "parl -v" without Module::Signature installed could delete all > files within the current directory. Oops. > > - On *nix systems, pp-generated executables erroneously linked to > libperl even if "pp -d" (--dependent) is not set. > > - Spurious =cut directives in source files is now handled gracefully > by PAR::Filter::PodStrip. > > Thanks! > /Autrijus/
