What should be in a README anyway (was Re: Phalanx / CPANTS / Kwalitee)
On Tue, Oct 14, 2003 at 01:14:44PM -0700, Randal L. Schwartz wrote: > Of course, if you have a well-written name/synopsis/author info, > I guess this is enough. But now we've just shifted the problem. Well when I initially thought about it I was answering the needs of someone who wanted to know: what the module does, what its api looks like, and what else is it going to want me to install. If there are other more tangible needs for a README that contribute to the magical "makes me want to untar it" factor then I'd like to accommodate them but first I'll need to know what they are. Can you offer pointers? -- Richard Clamp <[EMAIL PROTECTED]>
Re: Phalanx / CPANTS / Kwalitee
Hi!
On Mon, Oct 13, 2003 at 02:57:34AM -0700, Michael G Schwern wrote:
> On Mon, Oct 13, 2003 at 10:28:29AM +0200, Thomas Klausner wrote:
> > * contains files:
> > * Makefile.PL or Build.PL or configure
>
> configure?
there are currently 4 dists on CPAN that only include a configure script
(makepp-1.19, glist-0.9.17a10, swig1.1p5, shufflestat-0.0.3)
179 do not include any of Makefile.PL, Build.PL or configure.
Quite a lot come with two or three of those files.
But IMO a 'configure' is not a proper subsitute for Makefile.PL/Build.PL, so
I won't use it as a kwalitee indicator.
> > * t/* or test.pl
>
> test.pl is not a good thing. It doesn't get parsed by "make test". It just
> runs and spits the output to the screen. If a test fails, "make test" will
> still succeed. Still, its better than nothing at all.
So t/* should hint more kwalitee than only test.pl
> > Some ideas:
> > * wheater PREREQ_PM matches used modules (hard!)
>
> Nearly impossible since:
>
> require Foo::Bar if $some_condition;
There are obvioulsy edge cases, but one can detect the most plain
use/require statements.
> Otherwise Module::Info->modules_used() can handle it.
modules_used needs to compile the module, so it cannot be used for CPANTS.
--
#!/usr/bin/perl http://domm.zsi.at
for(ref bless{},just'another'perl'hacker){s-:+-$"-g&&print$_.$/}
Re: Phalanx / CPANTS / Kwalitee
Hi!
On Mon, Oct 13, 2003 at 07:46:02PM +0200, Tels wrote:
> > * POD/Code ratio (what would be a good measurement?)
>
> No. Some _very_ complex code takes little documentation like:
Agreed.
> Probably something like:
>
> *.pm file has more than 1000 lines of code => bad
>
> :-)
CPANTS does/will count lines of code (and POD). While those numbers probably
won't go into a kwalitee rating (while it still might be an interesting
metric, so I'm not sure..), anybody can grab them and do their own weighting.
("I /only/ use modules with less than 1000 lines of code. Everything longer
must be crap...")
--
#!/usr/bin/perl http://domm.zsi.at
for(ref bless{},just'another'perl'hacker){s-:+-$"-g&&print$_.$/}
Re: Phalanx / CPANTS / Kwalitee
Hi!
On Mon, Oct 13, 2003 at 05:53:15PM -, Rafael Garcia-Suarez wrote:
> > * POD/Code ratio (what would be a good measurement?)
>
> use Pod::Coverage ?
As far as I know, Pod::Coverage compiles the module, which makes it not
suitable for CPANTS.
OTOH, it might be an interesting idea to allow CPANTS to compile code (and
thus use stuff like Devel::Cover) if not testing the whole of CPAN, but
"linting" one distribution.
--
#!/usr/bin/perl http://domm.zsi.at
for(ref bless{},just'another'perl'hacker){s-:+-$"-g&&print$_.$/}
Re: Phalanx / CPANTS / Kwalitee
Thomas Klausner wrote: > there are currently 4 dists on CPAN that only include a configure script > (makepp-1.19, glist-0.9.17a10, swig1.1p5, shufflestat-0.0.3) > > 179 do not include any of Makefile.PL, Build.PL or configure. > > Quite a lot come with two or three of those files. Could we infer that a distribution that comes with several Makefile.PLs may have an overcomplicated build process, maybe indicating a low kwalitee ? See for example : http://search.cpan.org/~msergeant/XML-Parser-2.34/MANIFEST
Re: Phalanx / CPANTS / Kwalitee
On Wed, Oct 15, 2003 at 12:03:58PM +0200, Thomas Klausner wrote: > On Mon, Oct 13, 2003 at 05:53:15PM -, Rafael Garcia-Suarez wrote: > > use Pod::Coverage ? > > As far as I know, Pod::Coverage compiles the module, which makes it not > suitable for CPANTS. Afraid so. It's an eventual TODO-list item to steal/use the PPI parser from Claes' CPANXR rather than compiling the code. The compilation is there to answer the question "what methods does package X define". -- Richard Clamp <[EMAIL PROTECTED]>
Re: Phalanx / CPANTS / Kwalitee
Rafael Garcia-Suarez <[EMAIL PROTECTED]> writes: >Thomas Klausner wrote: >> there are currently 4 dists on CPAN that only include a configure script >> (makepp-1.19, glist-0.9.17a10, swig1.1p5, shufflestat-0.0.3) >> >> 179 do not include any of Makefile.PL, Build.PL or configure. >> >> Quite a lot come with two or three of those files. > >Could we infer that a distribution that comes with several Makefile.PLs >may have an overcomplicated build process, maybe indicating a low >kwalitee ? Should I infer that to get Tk's kwalitee up it should build as a one monolithic .so ? > >See for example : >http://search.cpan.org/~msergeant/XML-Parser-2.34/MANIFEST
Re: Phalanx / CPANTS / Kwalitee
Nick Ing-Simmons <[EMAIL PROTECTED]> wrote: > >Could we infer that a distribution that comes with several Makefile.PLs > >may have an overcomplicated build process, maybe indicating a low > >kwalitee ? > > Should I infer that to get Tk's kwalitee up it should build as a > one monolithic .so ? I don't know, I'm asking. So it's necessary to have one separate Makefile.PL by .so built ?
Re: Phalanx / CPANTS / Kwalitee
On Wednesday, Oct 15, 2003, at 11:09 Europe/London, Rafael Garcia-Suarez wrote: Thomas Klausner wrote: there are currently 4 dists on CPAN that only include a configure script (makepp-1.19, glist-0.9.17a10, swig1.1p5, shufflestat-0.0.3) 179 do not include any of Makefile.PL, Build.PL or configure. Quite a lot come with two or three of those files. Could we infer that a distribution that comes with several Makefile.PLs may have an overcomplicated build process, maybe indicating a low kwalitee ? I don't think so. For example I'm planning to release my modules with Build.PL and Makefile.PL in the future (because I like Module::Build, but want to continue to support people using CPAN). (maybe more than one == higher kwalitee :-) Adrian
Re: Phalanx / CPANTS / Kwalitee
Thomas Klausner sent the following bits through the ether: > OTOH, it might be an interesting idea to allow CPANTS to compile code (and > thus use stuff like Devel::Cover) if not testing the whole of CPAN, but > "linting" one distribution. Yes. We've been thinking about this. It either needs stealing buildd from Debian, having a box we don't mind destroying every so often, or having a VMware virtual machine we can undo easily. What we need is more free time ;-) Leon -- Leon Brocard.http://www.astray.com/ scribot.http://www.scribot.com/ ... Ethernet: A device for catching the Ether Bunny
Re: Phalanx / CPANTS / Kwalitee
Rafael Garcia-Suarez <[EMAIL PROTECTED]> writes: >Nick Ing-Simmons <[EMAIL PROTECTED]> wrote: >> >Could we infer that a distribution that comes with several Makefile.PLs >> >may have an overcomplicated build process, maybe indicating a low >> >kwalitee ? >> >> Should I infer that to get Tk's kwalitee up it should build as a >> one monolithic .so ? > >I don't know, I'm asking. >So it's necessary to have one separate Makefile.PL by .so built ? As it stands it needs even more rule-overriding hackery than Tk already does to build multiple .so/.dll per/Makefile. This by the way is the main reason I haven't yet done a Build.PL for Tk - I haven't figured out how to "build" sub-modules.
Re: Phalanx / CPANTS / Kwalitee
On Wed, Oct 15, 2003 at 02:10:46PM +0200, Rafael Garcia-Suarez wrote: > Nick Ing-Simmons <[EMAIL PROTECTED]> wrote: > > >Could we infer that a distribution that comes with several Makefile.PLs > > >may have an overcomplicated build process, maybe indicating a low > > >kwalitee ? > > > > Should I infer that to get Tk's kwalitee up it should build as a > > one monolithic .so ? > > I don't know, I'm asking. > So it's necessary to have one separate Makefile.PL by .so built ? I don't think you can or should try to infer anything useful from the number of Makefile.PL's. Tim.
Re: Phalanx / CPANTS / Kwalitee
Hi!
On Wed, Oct 15, 2003 at 03:24:22PM +0100, Leon Brocard wrote:
> Thomas Klausner sent the following bits through the ether:
>
> > OTOH, it might be an interesting idea to allow CPANTS to compile code (and
> > thus use stuff like Devel::Cover) if not testing the whole of CPAN, but
> > "linting" one distribution.
>
> Yes. We've been thinking about this. It either needs stealing buildd
> from Debian, having a box we don't mind destroying every so often, or
I see the biggest concern not so much with security / conflicting modules,
but with modules that interface to third party code and thus need it
installed (DBD::*, ImageMagick, GD, ...)
Or is some build tool able to automatically fetch needed non-perl libs?
--
#!/usr/bin/perl http://domm.zsi.at
for(ref bless{},just'another'perl'hacker){s-:+-$"-g&&print$_.$/}
Re: Phalanx / CPANTS / Kwalitee
On Wed, Oct 15, 2003 at 12:09:05PM +0200, Rafael Garcia-Suarez wrote: > Thomas Klausner wrote: > > there are currently 4 dists on CPAN that only include a configure script > > (makepp-1.19, glist-0.9.17a10, swig1.1p5, shufflestat-0.0.3) > > > > 179 do not include any of Makefile.PL, Build.PL or configure. > > > > Quite a lot come with two or three of those files. > > Could we infer that a distribution that comes with several Makefile.PLs > may have an overcomplicated build process, maybe indicating a low > kwalitee ? No, there's nothing wrong with having sub-build directories. -- Michael G Schwern[EMAIL PROTECTED] http://www.pobox.com/~schwern/ Kindly do not attempt to cloud the issue with facts.
Re: Phalanx / CPANTS / Kwalitee
On Wed, Oct 15, 2003 at 11:53:45AM +0200, Thomas Klausner wrote: > > configure? > > there are currently 4 dists on CPAN that only include a configure script > (makepp-1.19, glist-0.9.17a10, swig1.1p5, shufflestat-0.0.3) configure's an alien idiom on CPAN. If its found, I think it should be evaluated on a case-by-case basis. For example, looking at the above, makepp looks like its trying to emulate autoconf (sensible, I guess, since he's shipping a make replacement) and its doing it the right way. configure is a wrapper around a perl script that spits out a Makefile. Unfortunately, he shipped the generated Makefile. Still, it would have been better if configure was a wrapper around a Build.PL or Makefile.PL. Its been uploaded in the last few months. glist, OTOH, is shipping a configure written in shell. Sorta silly. He also shipped generated configure files that shouldn't be shipped (like config.cache). Since the dist is pure Perl, the guy is probably just comfortable with the configure/shell layout. It was touched last year. swig is a C program using autoconf, so that seems legit. It hasn't been touched since 1998. There's newer versions of Swig out there. It should probably be taken down. shufflestat is using what looks like a shell autoconf for a pure Perl program. Again, sorta silly. Its also shipping CVS directories and directories that would have been empty were it not for the empty CVS directory in it. :) This all suggests another check: stray files. Emacs backup files. CVS directories. Empty directories. #...# backup files. Makefiles shipped with Makefile.PL, Build and _build shipped with Build.PL, blib/... > > test.pl is not a good thing. It doesn't get parsed by "make test". It just > > runs and spits the output to the screen. If a test fails, "make test" will > > still succeed. Still, its better than nothing at all. > > So t/* should hint more kwalitee than only test.pl Yes. -- Michael G Schwern[EMAIL PROTECTED] http://www.pobox.com/~schwern/ Death follows me like a wee followey thing. -- Quakeman
Re: Phalanx / CPANTS / Kwalitee
Michael G Schwern wrote in perl.qa : > This all suggests another check: stray files. Emacs backup files. CVS > directories. Empty directories. #...# backup files. Makefiles shipped > with Makefile.PL, Build and _build shipped with Build.PL, blib/... In other words, the contents of the default MANIFEST.SKIP.
Re: Phalanx / CPANTS / Kwalitee
On Wed, Oct 15, 2003 at 10:34:19PM -, Rafael Garcia-Suarez wrote: > Michael G Schwern wrote in perl.qa : > > This all suggests another check: stray files. Emacs backup files. CVS > > directories. Empty directories. #...# backup files. Makefiles shipped > > with Makefile.PL, Build and _build shipped with Build.PL, blib/... > > In other words, the contents of the default MANIFEST.SKIP. Hey, yeah! -- Michael G Schwern[EMAIL PROTECTED] http://www.pobox.com/~schwern/ WOOHOO! I'm going to Disneyland! http://www.goats.com/archive/980805.html
Re: Phalanx / CPANTS / Kwalitee
> Yes. We've been thinking about this. It either needs stealing buildd > from Debian, having a box we don't mind destroying every so often, or > having a VMware virtual machine we can undo easily. What we need is > more free time ;-) > User Mode Linux (limiting to Linux, of course) might be a lighter weight way to do this. -R
