Stories
Slash Boxes
Comments
NOTE: use Perl; is on undef hiatus. You can read content, but you can't post it. More info will be forthcoming forthcomingly.

All the Perl that's Practical to Extract and Report

use Perl Log In

Log In

[ Create a new account ]

mpeters (5802)

mpeters
  (email not shown publicly)
http://petersfamily.org/
AOL IM: michael00peters (Add Buddy, Send Message)

Technorati Profile [technorati.com]

Journal of mpeters (5802)

Sunday March 29, 2009
08:45 AM

Module::Build::TAPArchive

TAP Archives are nice little packages of your test run that contain all the different TAP streams as well as test suite metadata. It's an easy way to store the results of a test run and then analyze them later. This is what Smolder uses but it's not specific to Smolder.

While it's easy to use TAP::Harness::Archive yourself to create your own harness it's kind of overblown for small projects. prove -a works, but sometimes you really want it as part of your standard perl Build.PL && ./Build test style of work.

Enter Module::Build::TAPArchive. Just use it instead of Module::Build in your Build.PL and then you can do

    perl Build.PL && ./Build test_archive

It's not doing anything really difficult behind the scenes, it should just make life a little easier for anyone who wants to play with TAP archives.

Thursday March 05, 2009
03:54 PM

eat your first cookie before asking for seconds

I like Gabor a lot and I think his work with Padre is really going to pay off in efforts to attract new blood to the Perl community, but I couldn't read the latest post on his blog without writing this response (he doesn't have comments on his blog, instead he encourages people to respond to him on their own blogs).

You should read his original post before continuing.

Asking TPF to fund 2 more grant ideas when you haven't even finished the last one they gave you is just asking to be rejected. It reminds me of my 5 year old asking for more cereal when he still has half his bowl left. I'm extremely unlikely to get up and get him some more.

I've been on the receiving end of the grant process and I know it's really hard to get motivated enough to finish those grants (the money doesn't help all that much). And in his case he admits to biting off a bit more than he can chew with the XS components. So I understand his desire to drop his current grant project.

But he should have done that prior to applying for more grants. And even then I wouldn't blame TPF for not accepting him for the next round if he didn't finish his last.

Like I said, Gabor is a great guy and really does good work, but I don't think he should see the rejection of his grants as a rejection of Padre by TPF. Instead look at it like they are good stewards watching over the Perl community's coffers and limited resources and being cautious with investments.
Friday June 13, 2008
01:49 PM

LD_LIBRARY_PATH trick PART 2

In my last post I talked about a trick with setting LD_LIBRARY_PATH in the same script that needs it. Then I ran into some problems with code being run by a perl -e scenario. My new solution (with some help from rjbs) is actually even simpler than what I'm doing now. The downside is it's not really portable. Should work for most *nixes and I only really need it to work on Linux, so I'm good.

The answer lies in asking /proc what command line options I was invoked with:

$ENV{LID_LIBRARY_PATH} = '/some/computed/path';
exec(split("\0", `cat /proc/$$/cmdline`));

Thursday May 15, 2008
06:05 PM

LD_LIBRARY_PATH trick

I have this problem a lot. I need to link some Perl module against some C library that is not in the standard system libs. Usually this is because it's a locally installed C thingy (like Swish-e or some payment processor API) that I don't want to pollute the system install. Plus it's nice and easy if the whole application is all bundled together.

The easy way to do this is to set the $ENV{LD_LIBRARY_PATH} from your shell to point to the right place. But this is a bother, and not to mention really hard to do if you need some dynamic bits to figure out what the right path is.

And for security reasons, programs can't set their own local version of LD_LIBRARY_PATH and have it be respected by the system's loader. It has to be done by the calling process (in most cases the shell).

This little trick let's a script set it's own LD_LIBRARY_PATH by relaunching the same program with the same arguments as a new sub-process. Simply put a BEGIN block like this in your script:

BEGIN {
    if(! $ENV{LD_LIBRARY_PATH} ) {
        $ENV{LD_LIBRARY_PATH} = join(':', @some_paths_I_need);
        my $self_cmd = "$^X $0 " . join(' ', @ARGV);
        system($self_cmd);
        exit;
    }
}

Monday April 07, 2008
05:02 PM

Oslo QA Workshop and Smolder

Smolder 1.21 has just been released thanks to the Oslo Perl QA Workshop and my
trip sponsored by Oslo.pm and Linpro (Linpro was an amazing host for this event. They gave us free reign of their offices for the whole weekend, free food, drinks and travel). This releases fixes several installation
bugs on various platforms and other small bugs. It also adds several new
features. The highlights are:

+ Project categories have now been replaced by the more flexible smoke report tags
+ TAP archives can now be downloaded through the UI for a given report
+ Graphs can now be done on test run duration if that info was provided in the
archive. This was inspired by Ovid's work on the BBC's test suite's run time.
+ The default DB-Platform is now SQLite instead of MySQL

See the changle log for full details.
http://sourceforge.net/projects/smolder

I was also pretty heavily involved in the group working on the next version of TAP. I have to say that this gave me a new appreciation for standards committees. It's really hard to balance extensibility, backwards compatability and ease of use. Especially with TAP since one of it's goals is being extremely human readable. We definitely made some progress and all of the new, slightly less readable features are completely optional so if authors don't want to use them, they don't have to. But Smolder is going to have fun with all this new information coming it's way!

Sunday May 20, 2007
11:31 AM

[smolder grant] progress update

I found some time this weekend to do some more work on my TPF grant to improve Smolder. I'll be doing a presentation to the Raliegh Perl Mongers on Smolder the first week of June, so I'll need to find some more time pretty soon too. Plus I'll need to definitely finish this round of work before YAPC and OSCON.

The first milestone (ditching the old XML format for plain TAP) is now feature complete (with tests of course). The accepted format is now a TAR archive containing TAP files with one YAML file with META information about the test run. To see more documentation see http://smolder.svn.sourceforge.net/viewvc/smolder/trunk/docs/upload_file_format. pod?view=markup.

The next 2 milestones have to do with setting up a system to easily automate CPAN style distributions for automatic testing from svn, etc. I've put this off for a few days but will get to it soon.

The 4th milestone was to add project specific RSS feeds. I was originally planning on using XML::RSS, but I've decided to go a simpler route. I don't need to parse RSS or do anything fancy other than emit XML using RSS tags. Using a template would be much easier. It also means I don't need to include the dependencies of XML::RSS inside of Smolder either, so it's that much smaller.

This milestone is now feature complete as well. That just leaves 2 left.

Saturday April 28, 2007
04:44 PM

[Smolder Grant Update] more work with TAP::Parser

I was a little delayed in my work this week ($work and fun stuff) because my SVK repository got really screwy and wouldn't let me push changes back for certain branches. So I took a whole lot of time troubleshooting that and then finally just deleting everything and starting my svk mirrors all over again.

After that I was able to jump right back into the gutting of the XML stuff from smolder in favor or plain ol' TAP. This went pretty smoothly and I was able to get a nice an intial implementation working, so I committed it back to the svn repo. I was a little pressed for time, so the test suite is still broken, but that shouldn't take too much time to get into order.

While working with the code that I haven't touched in a while, I noticed a lot of other areas (not related to my grant work) that could use some sprucing up too. Plus there's lot of new tricks and tips I've picked up from my work on other stuff (mostly the AJAXification of Krang) that would be useful for smolder to. But I guess that's the way it goes. There's always something to learn and never enough time to backport those learnings to earlier projects.

Monday April 23, 2007
08:50 AM

[Smolder Grant Update] XML = TAP

After battling some sickness and a little bit of yak shaving, I've finally started work on my grant for smolder.

The first deliverable is to replace the custom XML schema that I used (which was based on the internal representation of a test run used by Test::TAP::Model. I know, I know... this was a really bad thing to do, but so it goes (and yes, Test::TAP::Model has changed their internal representation). I want to replace the XML with straight TAP, but there are a few complications. TAP doesn't currently support multiple files. Each TAP "stream" correlates to one test file and the harness keeps track of the results to show the viewer at the end. Since I want smolder users to upload 1 file per test run I need something to contain them. I could wrap all those TAP streams in some container format (YAML, JSON) but then I'm no better of than the XML route I currently have.

So I've decided that smolder will take a simple archive file (tar/tar.gz initially) which contains the various TAP files. Since we sometimes want metrics like "How long did this test run take" or "What order where these files run in" (since it doesn't have to be alphabetical) then there will also be an optional YAML (or JSON I haven't decided) file in there with this meta information.

So the first step was ripping out all the XML stuff currently in Smolder and it's reliance on Test::TAP::Model and friends. Then to replace it with a TAP parser (using TAP::Parser). This is currently about half way done, and I hope to find the time to finish it this week.

Sunday April 01, 2007
10:37 AM

TPF Grant for Smolder and OSCon

I'm pretty excited that my TPF Grant for improving Smolder has been accepted. Some of the improvements are minor, but others are pretty big. I've been watching the progress of TAP::Parser with enthusiasm since it means I can get rid of the YAML/XML transport hack that Smolder currently uses to get test results into the server. Hopefully this will also give some evangelism for TAP as a nice, simple and really human readable format for testing.

The grant also comes at a good time because my proposal for OSCon this year was accepted and I'll be talking about an updated testing toolkit for web applications. So Smolder will be all nice and shiny from the grant work before the talk. Which will be a nice chance to show it off and hopefully get more people using it.

I'll be updating this journal pretty regularly as I work on this grant, so stay tuned if you're interested.

Monday September 25, 2006
01:37 PM

Smolder Slides

I gave the first presentation on smolder at the first Pittsburgh Perl Workshop on Saturday. It was a nice little conference that was well worth the money (and not just because it was free to speakers :)

In case others are interested and missed it, my slides are found here.