test failures during cpan installation

[Boris Glawe]

Hello,

I'd like to install the current version of bestpractical's request
tracker ([url]http://www.bestpractical.com/rt/[/url]), which is written in perl.
The installation skript wants me to install a lot of additional perl
modules.

Most of theses modules can be easily installed with the cpan command. In
cpan's command line interface, I enter something like "install Foo::Bar"

The problem is that some modules do not install due to some failing
tests. I get errors like "argument list too long" or "error, no reason
given", etc. I don't know how to solve such problems, since there's no
suitable error message, that tells me the reason for the failure.

I decided to install the modules with "force install Foo::Bar" which
succeeded. Can I expect that every will work?

In addition I have a module that cannot be installed due to linking
problems. It's the GD module. The linker ld says that it "cannot find
-lXpm", though the file libXpm.so.4 exists in /usr/X11R6/lib and the gcc
command has something like -L/usr/X11R6/lib in it's argument list.

rt is a very common request tracker system. I don't understand why the
installation is such a mess (sorry). A peace of software, that requires
C programming knowledge for installation and maintenance is not usable.

Can you help me with the usage of cpan? Is it "normal" that a lot of
modules can only be installed by forcing the installation, what went
wrong with the GD module?

Btw. my os is centos 4.3, which is technically identical to Redhat
Enterprice 4. My compiler is gcc 3.4.5 and the perl version is 5.8.5

thanks in advance

greets Boris

[Eric R. Meyers]

Hi Boris,

No your not crazy (yet), your mostly correct.

Boris Glawe wrote:

> Hello,
>
> I'd like to install the current version of bestpractical's request
> tracker ([url]http://www.bestpractical.com/rt/[/url]), which is written in perl.
> The installation skript wants me to install a lot of additional perl
> modules.
>

You just have to keep working through them all, like you're doing.

> Most of theses modules can be easily installed with the cpan command. In
> cpan's command line interface, I enter something like "install Foo::Bar"
>

cpanp (CPANPLUS) is also an option, but for either of these you should make
sure that you're configured to automatically follow the prerequisite
installations, so that everthing gets installed as needed.

> The problem is that some modules do not install due to some failing
> tests. I get errors like "argument list too long" or "error, no reason
> given", etc. I don't know how to solve such problems, since there's no
> suitable error message, that tells me the reason for the failure.
>

Some 'make test' failures are not uncommon, as long as it's not failing "too
much."

> I decided to install the modules with "force install Foo::Bar" which
> succeeded. Can I expect that every will work?
>

Excellent question, and you already know the answer, I'm sure. You seem
quite capable. No guarantee.

> In addition I have a module that cannot be installed due to linking
> problems. It's the GD module. The linker ld says that it "cannot find
> -lXpm", though the file libXpm.so.4 exists in /usr/X11R6/lib and the gcc
> command has something like -L/usr/X11R6/lib in it's argument list.
>

GD just installed for me, and I have FC4 linux, so look elsewhere.

> rt is a very common request tracker system. I don't understand why the
> installation is such a mess (sorry). A peace of software, that requires
> C programming knowledge for installation and maintenance is not usable.
>
> Can you help me with the usage of cpan? Is it "normal" that a lot of
> modules can only be installed by forcing the installation, what went
> wrong with the GD module?
>

If you can identify module problems, then check [url]http://rt.cpan.org/Public[/url]
for known bugs, or submit a new one.

> Btw. my os is centos 4.3, which is technically identical to Redhat
> Enterprice 4. My compiler is gcc 3.4.5 and the perl version is 5.8.5
>
> thanks in advance
>
> greets Boris

Write back, if you need to.

Eric

[Sherm Pendley]

"Eric R. Meyers" <ermeyers@adelphia.net> writes:

> Boris Glawe wrote:
>

>> The problem is that some modules do not install due to some failing
>> tests. I get errors like "argument list too long" or "error, no reason
>> given", etc. I don't know how to solve such problems, since there's no
>> suitable error message, that tells me the reason for the failure.
>>

> Some 'make test' failures are not uncommon, as long as it's not failing "too
> much."

A concrete example of that is the DBI::* modules, when you haven't supplied
the name of a test database and the login information. The self tests will
fail because they're unable to connect and run, but that doesn't indicate a
problem with the modules themselves.

Ultimately it's a judgement call - you need to review the error messages and
decide on a case-by-case basis if you think it's safe to them.

sherm--

--
Web Hosting by West Virginians, for West Virginians: [url]http://wv-www.net[/url]
Cocoa programming in Perl: [url]http://camelbones.sourceforge.net[/url]