Skip Menu |

This queue is for tickets about the Coro CPAN distribution.

Report information
The Basics
Id: 26341
Status: rejected
Priority: 0/
Queue: Coro

People
Owner: Nobody in particular
Requestors: dsteinbrunner [...] pobox.com
Cc:
AdminCc:

Bug Information
Severity: Normal
Broken in: 3.55
Fixed in: (no value)



Subject: Test pass even without prereq IO::AIO
I just installed Coro via the CPAN Shell and it tried to install IO::AIO but the test failed for it yet went on to install Coro. I'm not familiar enough with Coro to no how IO::AIO is used but since it is a prereq it would seem like the tests for the module would catch that it is not installed. Just thought I should point the occurrence out.
On Sat Apr 14 07:26:37 2007, dsteinbrunner@pobox.com wrote: Show quoted text
> I just installed Coro via the CPAN Shell and it tried to install > IO::AIO but the test failed for it
If you fail to install prerequisites you are completely on your own.
Subject: Re: [rt.cpan.org #26341] Test pass even without prereq IO::AIO
Date: Sat, 14 Apr 2007 11:02:33 -0400
To: <bug-Coro [...] rt.cpan.org>
From: David Steinbrunner <dsteinbrunner [...] pobox.com>
Marc_Lehmann via RT wrote: Show quoted text
> > <URL: http://rt.cpan.org/Ticket/Display.html?id=26341 > > > On Sat Apr 14 07:26:37 2007, dsteinbrunner@pobox.com wrote:
>> I just installed Coro via the CPAN Shell and it tried to install >> IO::AIO but the test failed for it
> > If you fail to install prerequisites you are completely on your own.
I'm not sure you understand, I did not fail to install IO::AIO but rather the tests for IO::AIO failed. I just so happened to catch that the tests failed but I can see someone else not noticing this when using the CPAN Shell and thinking that they had successfully installed Coro along with its dependencies. It seems odd that the tests for Coro could pass when one of its prerequisites is not present. From the outside looking in, it would appear more tests are needed to catch this situation. Or is it that IO::AIO is not that important to the functionality of Coro? I should note that my interest in Coro is due to it being a prerequisite of Continuity. Thanks, -- David Steinbrunner
On Sat Apr 14 11:02:53 2007, dsteinbrunner@pobox.com wrote: Show quoted text
> I'm not sure you understand,
Well, at least I am sure I understand: If you install perl, and it happens to have a bug, then you think that Coro should, in its test suite, check for the full integrity of perl. Because perl is a dependency of Coro, and Coro is responsible for testing that its prerequisite is installed correctly. Thats exactly your logic, just applied to perl and not IO::AIO. You cna apply your reasoning to your operating system, too, without any change. Sorry to say so, but that is idiotic. Coro should NOT contain the testuite for perl, and should NOT contain the testsuite for IO::AIO. Coro should only test that Coro works, and it might or might not do that very well, but that is a completely different issue. Show quoted text
> I did not fail to install IO::AIO
So you forced an install of IO::AIO despite the tests telling you there might be a problem. Sure, why not, that is your choice. As I said, you are completely on your own when you do that. Show quoted text
> failed but I can see someone else not noticing this when using the CPAN
The CPAN shell does not install a module when its testsuite fails unless you force it to. So if somebody else forces installation despite test failures he/she is also on his/her own. It is really simple. A user who doesn't care *will* see the testsuite failure of IO::AIO because CPAN will not install it. Show quoted text
> Shell and thinking that they had successfully installed Coro along
with its Show quoted text
> dependencies.
The he or she must be very dumb, forcing module installation without looking wether the module testsuite failed, or very knowledgable or desperate to install it despite knowing it fails, or simply not caring. All of that is a valid chocie for somebody, and Coro should not impede him or her on that. Show quoted text
> It seems odd that the tests for Coro could pass when one of its > prerequisites is not present.
First you claim the prerequisite is present, now you claim it is not present. This cannot be. Please make sure, before replying again, that what you are telling makes any sense. Show quoted text
> From the outside looking in, it would appear > more tests are needed to catch this situation.
No, of course not. Show quoted text
> Or is it that IO::AIO is not > that important to the functionality of Coro?
It is very important if that part is used. But Coro is simply not responsible for installing its prerequisites or testing them. Coro is, if at all, only responsible for testing its own code, anything else is madness. Show quoted text
> I should note that my interest in Coro is due to it being a
prerequisite of Show quoted text
> Continuity.
I fully support Coro being a prerequisite of Continuity. Please do not reply to the bugtracker anymore. Using it is a pain for me, and it should be reserved for real problems (actually I would prefer if people would not use it at all). If you want to correct your explanation of what happened, feel free to contact me (see the Coro manpage for my e-mail address). Thanks a lot!