Good that it uses zip but it doesn't look like there is a way to make read from the par archive without creating a intermediate tempfile. Whether I use perl -MPAR or parl it always unpacks to the TEMP folder. Is there anyway to prevent decompression to TEMP?
Regards. ---------------------------------- On Fri, 11 Jun 2010 13:18:55 +0200 [email protected] (Roderich Schupp) wrote: > On Fri, Jun 11, 2010 at 9:30 AM, Xaero <[email protected]> wrote: > > I was wondering if there was any other way to compress exe's genereated by > > PAR > > First of all, if you're generating an exe then we are talking about > PAR::Packer, not PAR. > An exe - as generated by pp - consists actually of a fixed part (i.e. > it's the same for any exe) > which is kind of "self extracting". The second part is a .par > (generated by PAR) > - it's actually a standard zip file. It contains your script and (most > of) the Perl modules > used by it (transitively). > > I think compression of the second part is fairly good, other compressors > have other tradeoffs, e.g. slightly better compression, but slower > extraction speed. > Also there is no other widely supported "compressed archive" standard other > than > zip archives. By "compressed archive" I mean that you can extract > inidividual members > _without_ having to uncompress the whole shebang (a gzipped tarball is a > counterexample). > > Now for the first part: it contains some core Perl modules > (essentially PAR and its > dependencies like Archive::Zip), a custom Perl interpreter and the > shared libperl.dll > (or such) that the interpreter is linked against. Neither of this > three parts is currently > compressed in any way. On Windows we're talking ~1.5 MB here, mostly for the > perl interpreter and the libperl.dll. Steffen and I independently did > some experiments > that show that one can roughly half the size by simply storing it > zipped (using zlib). > Of course the self extractor would need to link in statically the > uncompress methods > from zlib, that would add ~40 KB (on Windows) to the self extractor, so this > would be definitely a win. > > I have some experimental patches that can build the "self extractor" wit this > modification and have it produce the same stuff as the current version. > Since we can't count on that a static libz is available on the > platform where you > install PAR::Packer and pp (esp. on Windows), we need to > > - bundle the zlib source with PAR::Packer > > - modify PAR::Packer's build system to build a static zlib (if necessary) from > these sources > > The latter has absolutely nothing to do with Perl, so we can't expect > any help from > Perl's module build system (i.e. ExtUtils::MakeMaker) for it. Fortunately, > Compress::Raw::Zlib seems to have succeeded in doing this, so we might > just steal their solution. But it's still the part that will consume > the most tuits, > so don't hold your breadth. > > > I've tried testing the -z option but it looks like it make very little > > difference but the exe I'm testing it on is generated by pp -e "print > > q(hello!!)" command. Which makes me wonder if the -z switch only affect the > > .par part and not the entire exe? > > Correct, -z only affects the .par part which is rather small for the > "hello world" example :) > In general, the gain from -z 9 vs -z 6 (the default) is marginal. > > Cheers, Roderich
