-----BEGIN PGP SIGNED MESSAGE-----
Moin,
On Friday 16 July 2004 08:57, Randy W. Sims wrote:
Gabor Szabo wrote:
I can see from the testers page that Devel::Cover is supposed to work on Windows. Is there a ppd distribution of it somewhere so I can install it on ActivePerl without a compiler ?
Currently if I type
ppm install Devel::Cover
I get version 0.2 of Devel::Coverage. Not what I wanted.
It builds ok for me on one of my Windows machines, so I'm not sure why it's failing ActiveState's build system <http://ppm.activestate.com/>. The log message isn't that helpful. I know that their system basically goes through CPAN trying to build each module. Only those that build out of the box without intervention make it to the repository. I haven't really checked, but one common cause of failure, I would suspect, are dynamic dependencies, i.e. dependencies that are not specified to MakeMaker.
However, I really have to question their methods to try and build modules:
http://ppm.activestate.com/BuildStatus/5.8-linux/linux-5.8/Math-BigInt-BitVect-1.12.txt
The dependecies could be easily fullfilled (newer versions of BigInt are in the core and on CPAN!) but apparently for some reason it doesn't fullfill them, and then proceeds anyway, and the tests fall flat on their face (doh!).
I really wonder what is going on with their build system,
It looks like it is somehow failing to detect the failed dependencies. I don't know what versions of M::BI and M::BF ship with ActiveState's perl-their releases don't neccessarily conicide with official releases. I'm guessing based on observations that their script does something like
* Clean libs from previous installs, restoring the perl install to a clean slate.
* Download and untar the module.
* Run make.
* If there are unsatified dependencies, install them recursively.
* Run make test.
* If it passes, build binary ppm and move it to the repository.
The two places where it is likely to fail are 1) building & installing a dependency, and 2) doing the make test.
Your module is obvious failing during the 'make test'. It looks like it didn't detect the missing dependencies or it would have tried to install them first and if it couldn't it would have failed at that point...
But most of this is just guessing. I'd try posting to ActiveState's ppm mailing list <http://aspn.activestate.com/ASPN/Mail/>.
