Re: Re: Re: Re: Module::Build and the PPM

[autarch]

Yes. Bug reports and complaints were ignored. When the subject of releasing packages into CPAN _without_ a Makefile.PL came up the reaction was not pleasant.

Are you claiming you actually got flamed? I find this hard to believe. Unfortunately, the list archives only appear to be available via the horrid sourceforge archive system, so it's hard to find the thread(s?) you're referring to.

Regardless, I don't recall seeing many instances of people releasing modules to CPAN with only a Build.PL file and no Makefile.PL of any sort whatsoever, so it appears that the people who've decided to use Module::Build have decided that it's best to include both. Certainly, that's what I do with my modules.

Thanks for your comments autarch. With an attitude like that its no wonder you dont get the issues being raised here. Maybe if folks like yourself took the concerns of folks like us a little more seriously Module::Build would be in wider use, and recieve wider encouragement.

Maybe you shouldn't start by accusing the people working on it of being arrogant and criticizing them for things they have no control over (like someone uploading PPMs to CPAN!). That sort of stuff makes it a little harder to listen, since it's hard to find the substance amid the noise being generated.

For example, if you dislike the PPM- prefix used by default, maybe suggest something else (actually, Barbie did, and his suggestions seem reasonable). It's worth noting that there really doesn't appear to be a standard for this. ActiveState appears to use something like "Foo-Bar-0.01.zip" and Randy Kobes' repository uses "Foo-Bar.tar.gz". If Module::Build were to generate something like this, it would obviously be much more problematic than prefixing it with "PPM-".

But really, this whole thread starts from a basic misapprehension of Barbie's, which is that somehow that Module::Build (aka, its developers), by making PPM building easier, is therefore responsible when someone uploads it to CPAN. If PPMs don't belong on CPAN (and I'm thinking that they probably don't) then ask the person doing the uploading to stop. Problem solved.

[demerphq]

Are you claiming you actually got flamed?

Well I didnt rate it as a flaming. It was more like "yeah so what you dont like the decision. Thats the decision, its not going to change get over it." It was pretty clear that public opinion, or maintaining backward compatibility was not a priority. As for bug reports, i got tired of sending them out after the first couple were totally ignored. The impression I got was the attitude was like "we dont know who you are and we dont care what you think and oh btw getting it running on Win32 isn't a priority for us." Which essentially lost the efforts of at least one (somewhat) competent Win32 Perl Programmer who normally is inclined to help and to post patches.

Maybe you shouldn't start by accusing the people working on it of being arrogant

Well, I think their conduct and approach to Module::Build has been arrogant. I'm sorry if that upsets you, or that you feel that is unwarranted criticism, but the fact is that is the perception of a lot of people I've chatted with on this subject. Maybe what needs to happen is that the M::B crew actually listens to people, and perhaps even solicits their opinions and thoughts. If the ambition of this project is to change the working processes of tens of thousands of programmers dont you think a little community interfacing should occur?

Let me give you an example, ive seen you post a numer of "defend the module" posts. Have you ever posted a Meditation asking for the communities position on the changes being contemplated by M::B? Have you posted postion documents on CLPM, P5P and the other community locuses? Ive not seen any. Has anybody every asked the folks outside the M::B and CPANPLUS cliques how they feel about things? I bet not because I bet if it were put up to a vote there never would be another Build.pl uploaded to CPAN.

Basically the impression of M::B is: It produces distributions that are unbuildable from CPAN. It doesnt (or didnt) run on Win32. The people involved don't seem to listen unless your name is widely known in the community already.

If you are in a position to change these things then do so and youll see a lot of folks who dis M::B turn into Cheerleaders. Come and tell us we are wacko because we raise issues that are valid in our eyes and you wont win many converts.

Cheers, and respect.

[autarch]

Basically the impression of M::B is: It produces distributions that are unbuildable from CPAN. It doesnt (or didnt) run on Win32.

Hey, EU::MM doesn't run out of the box on Win32 either, because it requires nmake! And if a module includes XS, it doesn't really matter what build system you're using, because most Win32 folks don't have a compiler installed. Win32 Perl has always had more problems installing modules, thus the existence of PPM. The fact that Module::Build didn't work great on Win32 in the past (though I think things have changed) is simply because the lead developer doesn't use Win32. When Win32 folks like Randy Sims and others have stepped forward to help make it work, they've been welcomed.

The people involved don't seem to listen unless your name is widely known in the community already.

Uh, you keep saying stuff like this, but your only support for it appears to be that you made a suggestion on the list that wasn't accepted. I know Ken personally, and I've been on the M::B list, and I just don't buy this at all. Not every suggestion will be accepted. Things will be done that some people disagree with. That happens all the time. Do you agree with every decision made in regards to CPAN, or PAUSE, or the Perl core, and so on? Of course not. That's the nature of projects that impact a lot of people.

[demerphq]

Uh, you keep saying stuff like this, but your only support for it appears to be that you made a suggestion on the list that wasn't accepted.

No. I just happen to be one of many who think the same way. And we've all been disregarded. How many folks have you seen in this thread or the other M::B threads saying the same thing? How many people have said the same thing elsewhere? When that many people hold the same opinion, even if I dont agree with it I'll respect it. It seems that the M::B crew think otherwise. And IMO those who think this breakage is acceptable or even a good thing are in the minority.

Regarding your Win32 comments, they dont wash. I _have_ nmake on my PC (and its freely downloadable in at least two forms) I also _have_ a C/C++ compiler on my PC (and a free version of cl.exe has been available from MS for some time now.) So im one of those folks who can get just about any module that doesnt have Linux specific library issues to install. EU::MM runs out of the box on my box, why shouldnt M::B? And if somebody files bug reports on a system that you dont directly work with dont you think it should be followed up on? I know I do.

Ultimately I hope M::B works out. But I think part of that will involve backing off from some of the decisions made, and possibly doing some community outreach. As it is M::B has a bad rep basically for a small set of reasons and until someone decides remove those causes I suspect itll stay that way.

Oh, and as a last thought has it occured to you what might happen? Folks might get so sick of M::B incompatibilities that they start parallel maintaining normal distros. How would you feel if someone started redistributing your work under a new name just becuase you decided to use M::B? To be honest I can see it happening. And that would just make the current EU::MM situation worse wouldnt it?