develooper Front page | perl.perl5.porters | Postings from September 2011

Re: A summary of Chip patches and potential impact

Thread Previous | Thread Next
Steffen Mueller
September 3, 2011 03:11
Re: A summary of Chip patches and potential impact
Message ID:
On 09/02/2011 10:49 PM, Reverend Chip wrote:
> To accompany Jesse's efforts to explain what makes a patch acceptable
> -- and I really like the phrase "free as in puppies" -- I wrote the
> below summary of the changes I have pending.  Steffen++ is CPAN
> smoking the magic flags patch; some of the other patches need CPAN
> smoking too.

Let me give you some update on the CPAN smoking. I've chosen to launch a 
full minicpan at both a recent bleadperl and the magicflags1 branch 
(well, the snapshot of that I took a while ago -- not sure if it's 
changed since). I'm smoking both on the same quad-core machine at the 
same time, one smoking process each. It's been running for... quite a 
bit over a week now, I think.

model name      : AMD Opteron(tm) Processor 285
cpu MHz         : 2600.000

$ wc -l cpan_perl_branch_smoke/alldist.txt
23293 cpan_perl_branch_smoke/alldist.txt

So 2 * 23293 distributions to smoke.

$ ll /scratch/steffen/smoke_out/perl-*/* | wc -l

That means from both perls combined, 16789 reports are in. This is going 
to take quite some time to finish.

But that's not all: I had a 25GB scratch directory created specifically 
for this on the machine by Dennis (thank you!). It currently looks like 

                        25G   19G  4.7G  81% /scratch/steffen

Which means that it won't finish the whole run without running out of 
disk space. This is something that could be conceptually solved in two ways:

- Hack the smoking software / CPANPLUS to clean up the build directory 
occasionally. It does not do that now. Thus the disk hog.
- Make my ugly smoker script re-startable.

I am the wrong person to do the former. The latter is just a SMOP, but I 
haven't had either inclination nor time to do it. I guess I'll have to. 
Yay. Not just that, I'll also have to hack it so that it can inspect a 
report output directory to see what's done since I can't go back in 
time. Yay me for planning ahead! :(

My hacks can be found on github at with some step-by-step 
instructions in todo.txt. It includes a slightly modified version of the 
comparison script by David Golden.

This leaves the fact that the CPAN smoking is just too damn slow to be 
throwing it at a lot of branches in the future. I do think it's viable 
for deep changes such as yours, but we need to be very careful with 
adopting a "if in doubt, let's CPAN smoke it" policy both for 
conservation of volunteer time, CPU time, and latency (this remark is 
for Jesse primarily).

Best regards,

Thread Previous | Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About