Bug #56117 for Module-Build: Refactor ACTION_diff and break out test libraries

Tue Mar 30 13:42:26 2010 SILASMONK [...] cpan.org - Ticket created

Subject:

Surprisingly limited support for subclassing

1.) It would be nice if the MBTest and DistGen modules could be published and better documented as they are very useful in testing subclasses of Module::Build. DistGen in particular has Module::Build hardcoded in a number of places and would need to be generalized. 2.) The ACTION_diff function seems surprisingly hardcoded. It would be better if it went through the various install_types and ran diff_${type}_files on each type.

Tue Mar 30 14:44:36 2010 dagolden [...] cpan.org - Correspondence added

Good points. You're welcome to just copy them into subclass distributions, of course. DistGen is much better documented than it was. -- David

Tue Mar 30 14:44:37 2010 The RT System itself - Status changed from 'new' to 'open'

Tue Mar 30 14:45:23 2010 dagolden [...] cpan.org - Subject changed from 'Surprisingly limited support for subclassing' to 'Refactor ACTION_diff and break out test libraries'

Tue Mar 30 14:45:23 2010 dagolden [...] cpan.org - Severity Wishlist added

Tue Mar 30 15:10:49 2010 SILASMONK [...] cpan.org - Correspondence added

On Tue Mar 30 14:44:36 2010, DAGOLDEN wrote: Show quoted text

> Good points. You're welcome to just copy them into subclass > distributions, of course. DistGen is much better documented than it was. > > -- David

Yeah I have copied them. I started off writing my own but realized I ought to use them. The hardcoding in DistGen required some manual editing however. Also I am very puzzled about the where ACTION_diff is written. It seems fragile and totally different from say ACTION_code. Thus subclassing seems very likely to break ACTION_diff. I also have another issue: 3.) If you to try in a subclass to remove the 'arch' build element you get into issues. I got round this by redefining 'install_types' to not return any type without a destination.

Tue Mar 30 15:45:17 2010 kwilliams [...] cpan.org - Correspondence added

On Tue Mar 30 15:10:49 2010, SILASMONK wrote: Show quoted text

> Also I am very puzzled about the where ACTION_diff is > written. It seems fragile and totally different from say ACTION_code. > Thus subclassing seems very likely to break ACTION_diff.

Yeah, ACTION_diff probably hasn't seen a lot of action recently. I wrote it ages ago (it was actually one of the "killer apps" in my mind for Module::Build) but I'm not sure many people have used it. It could probably use a fresh coat of paint or two.

Tue Mar 30 17:18:16 2010 SILASMONK [...] cpan.org - Correspondence added

On Tue Mar 30 15:45:17 2010, KWILLIAMS wrote: Show quoted text

> On Tue Mar 30 15:10:49 2010, SILASMONK wrote:

> > Also I am very puzzled about the where ACTION_diff is > > written. It seems fragile and totally different from say ACTION_code. > > Thus subclassing seems very likely to break ACTION_diff.

> > Yeah, ACTION_diff probably hasn't seen a lot of action recently. I > wrote it ages ago (it was actually one of the "killer apps" in my mind > for Module::Build) but I'm not sure many people have used it. It could > probably use a fresh coat of paint or two.

Related to ACTION_diff I am also looking to write an ACTION_sync that copies from the install destinations back to the source. I am not sure if that is of general applicability, but my subclass is basically supposed to be a deployment system, and sometimes you want to capture whatever is in prod.