| [reply] |
- Bad things for users
- When an installation fails for the user they have no idea whether this is their problem alone or a systematic problem with the dist
- When a newer version is released the user is unaware of any problems there might be with it before downloading/installing
- If there is a systematic problem with the current version the user has no reference as to which other versions are problem-free
- Bad things for authors
- No way to tell if a manual bug report received for any given platform is unique to that user
- No way to tell if a just-released dist has a glaring error (until hopefully manually notified by someone much later)
- Having to go back-and-forth with a bug reporter to get case info rather than having the full smoke tester failure available at a single click
- No way to tell if a dependency of their code has a usable current version or if the last n versions are all usable, etc.
- Bad things for perl
- Another important piece of infrastructure gone making it harder than it used to be for everyone to use the language (see also: AnnoCPAN, CPAN Ratings, etc.)
| [reply] |
- It will be the module author's job to test everything in all the environments.
- Great shame for Perl.
map{substr$_->[0],$_->[1]||0,1}[\*||{},3],[[]],[ref qr-1,-,-1],[{}],[sub{}^*ARGV,3]
| [reply]
[d/l] |
It will be the module author's job to test everything in all the environments.
I can't imagine any author doing that - I certainly won't be.
Mostly, I just test locally on Windows, Ubuntu, and freebsd - and hope for the best.
The thing that usually brings me undone is that I tend to forget that some people like to run old, buggy stuff - be it perl itself, or a prerequisite perl module, or a prerequisite C library.
Such oversights generally go undetected until the "right" testing environment turns up.
Recently, I wrote a test script for Math::GMPz that required Math::MPFR.
Naturally, the availability of Math::MPFR was first specified .... but I failed to also specify that the version of Math::MPFR had to be less than 18 releases out of date. That is, current version is 4.25, 4.07 is minimum required :-(
Without cpantesters, my life wouldn't change much.
I'd still be simply sitting back, fixing any issues that are brought to my attention.
The main change that the demise of cpantesters would make for me is that there would be fewer such issues brought to my attention.
Cheers,
Rob
| [reply] |
I concur with what ++choroba wrote in answer to
"... when CPAN testers ceases to exist?".
The ending of EPO does not necessarily mean the end of CPAN Testers.
A better question might have been "... if CPAN testers ceases to exist?".
| [reply] |
