Greetings.

While I have worked in Perl a long time, I am not the wizard to help
with abi-dumper.

I had been a long time Debian user until about April of 2014. Systemd
made me change to Gentoo, which is taking some time to get used to.

I would like to raise some points with respect to Perl packages. I can
accept that Debian had a policy that all packages have lowercase letter
names.

That policy does allow for name collisions more often than mixed case
package names. The first real name collision I had to decide upon
myself, was a Solaris installation, where I had Arc/Info installed, as
well as FSF Info pages. Two programs named "info" was clumsy. I think
mixed case names makes sense in some circumstances.

All packages which come in from some other repository (CPAN, CTAN,
R, ...) should be searchable by the repository, using terms from the
parent repository. The description of a package should display the
parent repository name.

Second, all CPAN test files and example files should come along with
the Debuan packge. Sometimes there is just not enough documentation,
and being able to see the test files or example files will suddenly
explain something better.

Looking at libxml-libxml-perl, we see that its description does mention
that it is a Perl interface that is named XML::LibXML. I suppose most
of us looking for Perl packages will know that its CPAN name is
XML::LibXML. But I have met people trying to install other packages,
who have gotten lost on similar issues.

It is nice that CPAN would like for all packages to have test cases.
From what I have seen, all of these test cases nominally assume it is
possible to always calculate exact answers (integer arithmetic). A few
times I have thought about how one might do testing of floating point
modules. I may be able to make some time to look into this again.
Perhaps a generic solution using Number::WithError is possible?

In the instance where some kind of package is an interface to a
library, it would probably be useful to say if this is SWIG, XS or
something else. In the case of Perl modules, is it using one of the
Inline packages, or is it using the full XS machinery?

Are there items in the POD man page, which should make it to the Devuan
package description? I am specifically thinking of Depends, but some
times these man pages reference URLs of interest.

Is Wikipedia becoming broad enough in scope, that man pages should
reference Wikipedia if possible?

Can that mechanism work with more newby friendly documentation? An
extension to man (or info).

History: I've been number crunching since 1979, mostly in FORTRAN, C
and Perl. I've dabbled in sysadmin and net admin, including doing a
partial port of Perl 4.x to QNX 2.x way back when.

I will guess some of the above is contrary to being minimal.


Dak? Are we supposed to learn from:
https://tracker.debian.org/pkg/dak

Have a great day!

Gord