Discussion:
[Pdl-porters] PDL::IO::Pic related error in testing while use PDL
Ingo Schmid
2014-03-17 10:44:13 UTC
Permalink
Hi,

I have a lot of test failures with pod-coverage.t in PDL::Dims.

https://rt.cpan.org/Ticket/Display.html?id=93527

Thanks to helpful testers it appears to be related to PDL::IO::Pic, so I
thought I post here.

Ingo
Chris Marshall
2014-03-17 13:04:06 UTC
Permalink
Hi Ingo-

A couple of warnings at the very top:

* A smart match one which I don't know about
except that smart match was introduced after
perl 5.8.x (I think) so the PDL::Dims would not
be usable for any "old perl" users.

* splice ref warning is odd, I would be sure that
nothing funny is going on there. There could
be issues related to new perl features that
need special mods to support for compatibility.

* If PDL::IO::Pic is a problem, then don't import
it for your module tests. At this stage in the
game I don't see a lot of developers interested
in retrofitting a mature and stable module for -T

In my experience, if you want no test failures,
it requires a combination of good code *and*
tests that are written to work.

Good luck,
Chris
Post by Ingo Schmid
Hi,
I have a lot of test failures with pod-coverage.t in PDL::Dims.
https://rt.cpan.org/Ticket/Display.html?id=93527
Thanks to helpful testers it appears to be related to PDL::IO::Pic, so I
thought I post here.
Ingo
_______________________________________________
PDL-porters mailing list
http://mailman.jach.hawaii.edu/mailman/listinfo/pdl-porters
Ingo Schmid
2014-03-17 13:50:32 UTC
Permalink
Hi Chris,

thanks for responding. I've uploaded a fixed version (i.e. no -T switch).
Post by Chris Marshall
Hi Ingo-
* A smart match one which I don't know about
except that smart match was introduced after
perl 5.8.x (I think) so the PDL::Dims would not
be usable for any "old perl" users.
The required minimum perl version is 5.12. I probably could rework it
for older versions, but not at the moment.
Post by Chris Marshall
* splice ref warning is odd, I would be sure that
nothing funny is going on there. There could
be issues related to new perl features that
need special mods to support for compatibility.
* If PDL::IO::Pic is a problem, then don't import
it for your module tests. At this stage in the
game I don't see a lot of developers interested
in retrofitting a mature and stable module for -T
What are you trying to say, exactly? I still have no idea what the issue
with PDL::IO::Pic is about and therefore sent it to the list.

I used module-starter to initialize the module. the pod-coverage.t was
created automatically with perl -T. I have no idea why, what it does
exactly, or if that makes sense, to be honest. I was looking for a
minimalistic way of creating a distribution since if PDL is installed,
it should and will just work, barring bugs ;).
Post by Chris Marshall
In my experience, if you want no test failures,
it requires a combination of good code *and*
tests that are written to work.
Good luck,
Chris
Post by Ingo Schmid
Hi,
I have a lot of test failures with pod-coverage.t in PDL::Dims.
https://rt.cpan.org/Ticket/Display.html?id=93527
Thanks to helpful testers it appears to be related to PDL::IO::Pic, so I
thought I post here.
Ingo
_______________________________________________
PDL-porters mailing list
http://mailman.jach.hawaii.edu/mailman/listinfo/pdl-porters
Chris Marshall
2014-03-17 14:31:20 UTC
Permalink
Post by Ingo Schmid
Hi Chris,
thanks for responding. I've uploaded a fixed version (i.e. no -T switch).
Post by Chris Marshall
Hi Ingo-
* A smart match one which I don't know about
except that smart match was introduced after
perl 5.8.x (I think) so the PDL::Dims would not
be usable for any "old perl" users.
The required minimum perl version is 5.12. I probably could rework it
for older versions, but not at the moment.
Post by Chris Marshall
* splice ref warning is odd, I would be sure that
nothing funny is going on there. There could
be issues related to new perl features that
need special mods to support for compatibility.
* If PDL::IO::Pic is a problem, then don't import
it for your module tests. At this stage in the
game I don't see a lot of developers interested
in retrofitting a mature and stable module for -T
What are you trying to say, exactly? I still have no idea what the issue
with PDL::IO::Pic is about and therefore sent it to the list.
Taint checking is to prevent unplanned access from
"outside" the perl program to the "inside" and requires
special handling of anything coming from the outside.

PDL::IO::Pic being an IO module is always reading and
writing from the "outside" world to the "inside perl" world.
As such is difficult if not impossible to truly get taint-safe.

One of the on-going thrusts for PDL is to make it fully
cross platform and system() calls and their ilk are very
non-portable across different systems.
Post by Ingo Schmid
I used module-starter to initialize the module. the pod-coverage.t was
created automatically with perl -T. I have no idea why, what it does
exactly, or if that makes sense, to be honest. I was looking for a
minimalistic way of creating a distribution since if PDL is installed,
it should and will just work, barring bugs ;).
If you wanted to add a feature request ticket to determine
which modules of PDL are taint-safe and have that documented,
that might be of interest to some users.
Post by Ingo Schmid
Post by Chris Marshall
In my experience, if you want no test failures,
it requires a combination of good code *and*
tests that are written to work.
Good luck,
Chris
Post by Ingo Schmid
Hi,
I have a lot of test failures with pod-coverage.t in PDL::Dims.
https://rt.cpan.org/Ticket/Display.html?id=93527
Thanks to helpful testers it appears to be related to PDL::IO::Pic, so I
thought I post here.
Ingo
_______________________________________________
PDL-porters mailing list
http://mailman.jach.hawaii.edu/mailman/listinfo/pdl-porters
Loading...