Andy Armstrong (a...@hexten.net)

User Information

Display Name:

Andy Armstrong

Partial Email Address:

a...@hexten.net

Posts:

25	total
3	in mod_perl
22	in Perl 5 Porters

5 Most Recent

All Posts

1) Andy Armstrong Re: [PATCH] Test::Harness should clean up after itself

+1 vote

Yeah, it's a bug in how we generate the filename for spooled TAP. We build the spool name from the...

Perl 5 Porters

[ Profile | Reply to group ]

[ Flat Thread Threaded ]

On 14 Nov 2008, at 11:31, Nicholas Clark wrote:
> Andy knew the cause, but I'm not sure if this is the best fix. Yes,  
> it will
> clean up, but the problem is that part of TH3 is writing  
> intermediate files
> outside of the directory tree that it's "supposed" to be using,  
> because the
> test names start with ../
>
> So it seems that it needs a better plan to cope with "interesting"  
> test names,
> with which cleanup will fall out naturally.

Yeah, it's a bug in how we generate the filename for spooled TAP. We  
build the spool name from the test name and if the test name  
includes .. we honour that. Oops.

My quick attempt at a fix didn't do the trick - for reasons as yet  
unknown. We're in the middle of a big push at work right now and I  
haven't had a chance to work out why my fix didn't fix. I'll make some  
time to prod it over the weekend.

Thanks for the patch in any case Chip :)

-- 
Andy Armstrong, Hexten

2) Andy Armstrong Re: ANNOUNCE Test::More/Simple/Builder 0.84

+1 vote

I was just about to ask about that. Still, I now have Mouse on all the machines I've upgraded -...

Perl 5 Porters

[ Profile | Reply to group ]

[ Flat Thread Threaded ]

On 15 Oct 2008, at 14:09, Michael G Schwern wrote:
> 0.82 accidentally shipped with a dependency on Mouse.  We're only  
> using that
> for experiments.


I was just about to ask about that. Still, I now have Mouse on all the  
machines I've upgraded - which is a good thing :)

-- 
Andy Armstrong, Hexten

3) Andy Armstrong Re: make distclean failures

+1 vote

Ah... :) I'll fix those tonight.

Perl 5 Porters

[ Profile | Reply to group ]

[ Flat Thread Threaded ]

On 8 Oct 2008, at 08:24, Rafael Garcia-Suarez wrote:
>> This suggests that it is the case.
>
> That's correct, but don't forget the other makefiles :


Ah... :)

I'll fix those tonight.

-- 
Andy Armstrong, Hexten

4) Andy Armstrong Re: make distclean failures

+1 vote

Ah. I see. Well the generated Makefile for Test::Harness looks odd to me - seems to assume the...

Perl 5 Porters

[ Profile | Reply to group ]

[ Flat Thread Threaded ]

On 7 Oct 2008, at 19:29, Nicholas Clark wrote:
>> I'm having a look at it now. I'm wondering why the Test::Harness  
>> stuff
>> starts out in ext. I'm assuming that modules in ext generally have an
>> XS component that needs to be compiled? It looks to me as if
>> Test::Harness should really be in lib - but I'm probably
>> misunderstanding something.
>
> The long term plan (I think) was to move most dual life modules to  
> ext, so
> that it's easier to update them from CPAN (or upstream's version  
> control)
> I moved Test::Harness there in blead already, because it made it way  
> easier
> to use the svn HEAD directly when profiling parallel testing of the  
> core.

Ah. I see. Well the generated Makefile for Test::Harness looks odd to  
me - seems to assume the source files are in ../../../lib - but I've  
only just started looking and don't really understand what it's trying  
to do yet :)

-- 
Andy Armstrong, Hexten

5) Andy Armstrong Re: make distclean failures

+1 vote

I'm having a look at it now. I'm wondering why the Test::Harness stuff starts out in ext. I'm...

Perl 5 Porters

[ Profile | Reply to group ]

[ Flat Thread Threaded ]

On 7 Oct 2008, at 17:07, Nicholas Clark wrote:
> Currently in blead after make distclean the following remain:
>
> ext/Test/Harness/t/ext
> ext/Test/Harness/t/ext/Test
> ext/Test/Harness/t/ext/Test/Harness
> ext/Test/Harness/t/ext/Test/Harness/t
> ext/Test/Harness/t/ext/Test/Harness/t/source_tests
> ext/Test/Harness/t/ext/Test/Harness/t/source_tests/harness
> lib/App
> lib/App/Prove
> lib/App/Prove/State
> lib/App/Prove/State/Result
> lib/Module/Pluggable
> lib/TAP
> lib/TAP/Formatter
> lib/TAP/Formatter/Console
> lib/TAP/Parser
> lib/TAP/Parser/Iterator
> lib/TAP/Parser/Result
> lib/TAP/Parser/Scheduler
> lib/TAP/Parser/Source
> lib/TAP/Parser/YAMLish
> lib/unicore/uni.data
>
>
> I know we've had detritus left before. What's the usual place that  
> we fix this?
> (Also, that ext/Test/Harness/t/ext etc looks suspicious. Anyone have  
> the time
> to figure out where that is coming from?)


I'm having a look at it now. I'm wondering why the Test::Harness stuff  
starts out in ext. I'm assuming that modules in ext generally have an  
XS component that needs to be compiled? It looks to me as if  
Test::Harness should really be in lib - but I'm probably  
misunderstanding something.

-- 
Andy Armstrong, Hexten

Highlight

Profile | Posts (25)

Home > People > Andy Armstrong