Re: [MacPerl] CPAN

[Matthias Ulrich Neeracher <neeri@iis.ee.ethz.ch>]

pudge@pobox.com (Chris Nandor) writes:
>At 8:52 1/15/97, Paul J. Schinder wrote:
>>I take it that Andreas software assumes a .tar.gz, or does it accept other
>>formats?  How smart is it?  It seems to me the simplest solution would be
>>for you or anyone else wanting a MacPerl package in CPAN to create a two
>>item .tar.gz, a README and the .sit.hqx.  His receiving software could be
>>modified to put the .sit.hqx (we *really* should be using .sit.bin) and the
>>README uncompressed into CPAN rather than the .tar.gz file itself.

Another possibility just occurred to me: We could simply drop back to stuffit
1.5.1. However, there is probably no archive left in the world which
distributes 1.5.1.

>Interesting.
>
>1.)  No, we should use .hqx.  .bin is too unreliable.  I know it is
>smaller, but too many people have had problems with it for me to condone
>its use.  Just my opinion.

I disagree. Most ftp clients handle MacBinary automatically and to folks like
me who are sitting on a 14400 link at home, 30% longer download times are not
a trivial disadvantage (Not to speak of 30% more archive disk space).

Furthermore, MacPerl itself has for several years been distributed as
MacBinary, and complaints have decreased over time.

>2.)  But if we are going to bother to .tar.gz it anyway, why create the
>.sit.hqx?  Almost anyone who uses MacPerl can get at .tar.gz files, simply
>because you need it to get at everything else in CPAN.

tar files have no concept of file types, so shared libraries and droplets
are going to break. One would have to do MacBinary encoding of the items
in the tar archive. It's probably better to only put text files in the
.tar, that still allows search engines and such to operate.

Matthias

-----
Matthias Neeracher   <neeri@iis.ee.ethz.ch>   http://www.iis.ee.ethz.ch/~neeri
   "One fine day in my odd past..." -- Pixies, _Planet of Sound_