home about news mgmt


2017-08-14 k <andk@cpan.org>
  • The beforemaintrelease page needs to go on a diet, it consumes too much of analysis' resources. I'll turn the production off for a while. Let me know if you are a regular consumer and missing it. I'm sure it can be improved but if nobody is missing it, we probably won't.

2016-04-12 k <andk@cpan.org>
  • I just removed the beta tag from the front page. 6 years of beta is enough and probably misleading.

2015-09-14 k <andk@cpan.org>
  • Analysis was stalled from 2015-08-15 to 2015-08-29 and again 2015-08-30 to 2015-09-14 due to to upstream data reconstruction.

2015-05-15 k <andk@cpan.org>
  • Analysis was stalled from 2015-05-03 to 2015-05-15 due to a major reconstruction of cpantesters.

2014-09-21 k <andk@cpan.org>
2014-06-28 k <andk@cpan.org>
  • These days I added a small ajax based layer to the /solved/ page, so that a status message appears below the headline when something relevant to this distro happens on the server. It works like this:

    If a new calculation is scheduled, it is noted as "...is scheduled". The message changes to "...has commenced" when the new calculation has started to run. And when the new calculation has finished, the text changes to a recommendation to reload the page.

    Also the number of Passes and Fails is updated regularly. When one of them increases, then a button appears that lets the user schedule a new calculation.

2014-06-26 k <andk@cpan.org>
  • At the Perl Mongers meeting yesterday, Slaven asked for the (then missing) page http://analysis.cpantesters.org/solved?distv=MarekPodHtml-0.49

    It was missing because the release was 2003, i.e. older than eight years. Since all the optimizations of the least six months (Postgres, Redis) have resolved enough bottlenecks, I simply bumped the maximum time we track releases from 8 to 16 years, and there we are. 16 years ago, that was the time when BJEPS/Data-Xtab-1.01.tar.gz was released (Hi Brian). We'll see when we will have to double the period next time. Let me know!

2014-05-26 k <andk@cpan.org>
2014-01-05 k <andk@cpan.org>
  • On perl5-porters I just announced the experimental new page http://analysis.cpantesters.org/beforemaintrelease

    It demonstrates a way how the cpantesters and analysis data can be consumed from a different perspective, namely for focussing QA on a new perl maint release.

  • For the new table I thought it would be a good idea to upgrade the jquery tablesorter. So there it is, both for the new and the old pages. It made the tables uglier and they more often sort badly, but at least it has the arrows in the sortable table headers.

  • Over the holidays analysis has got a redis based job queue that allows us to conveniently run regression calculations in parallel. I ♥ Redis.

2013-12-13 Andreas J Koenig <andk@cpan.org>
2013-12-09 Andreas J Koenig <andk@cpan.org>
  • Today we have finished migration of all database tables from sqlite to postgresql. Whatever went wrong with sqlite, we're now better in terms of both speed and concurrency, allowing us to make headway.

2013-11-29 Andreas J Koenig <andk@cpan.org>
  • At the London Perl Workshop Barbie talks about the amazingly improved speed of cpantesters. It looks like we could now change gears on analysis too to provide fresh statistical results faster than in the past. We are already a bit faster automatically now but there's still work left to get on par with the speed that cpantesters now support.

    Maybe we can soon remove all the *lt-matrix* links because the normal matrix is usually fresh enough.

2012-11-06 Andreas J Koenig <andk@cpan.org>
  • We've been low on fresh data recently while September and October are being re-scanned for missing reports. This has lead to the improvised cpantesters matrix site at and analysis now has a link labeled *lt-matrix* (for log.txt-matrix) on the homepage to this site until cpantesters is uptodate again.

2012-05-28 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
2011-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
  • Analysis was stalled between 2011-11-11 and 2011-11-21. The database exported by cpantesters was corrupted so analysis did not get fresh data.

2011-10-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
  • Analysis was stalled between 2011-08-27 and 2011-10-12 due to the reconstruction of cpantesters. Today we're trying to go into business again. One schema change has been identified and our code adjusted accordingly. We have no other way to verify our code than to simply run it. Please be sceptic, do not trust any results. Verify everything before you draw conclusions.

2011-02-13 Andreas J. Koenig <andk@cpan.org>
  • The porting to the virtual box went smoothely and analysis has taken up the full service again. Pause should now be as fast as it was before the analysis project started, which was pretty fast.

    If you're still facing a slow pause.perl.org please let me know. Then we must investigate somewhere else.

2011-02-12 Andreas J. Koenig <andk@cpan.org>
  • Complaints about PAUSE being slow have been voiced repeatedly. Since analysis.cpantesters.org is (was) running on the same box as pause.perl.org I'm starting an emergency plan for analysis.

    The sponsors have provided a virtual box that I can use for analysis for further explorations. I'm starting to port analysis to the other box today.

    If it turns out that analysis was to blame for pause slowness we'll keep the two split. I'll try to keep up the service as much as possible.

    Until further notice please expect reduced availability.

2011-01-23 Andreas J. Koenig <andk@cpan.org>
2011-01-01 Andreas J. Koenig <andk@cpan.org>
2010-08-14 Andreas J. Koenig <andk@cpan.org>
  • Analysis has survived the first week after the rewrite for Cpan Testers 2.0. It is slow and needs further work but it seems stable enough to invite users to make active use of it.

2010-08-07 Andreas J. Koenig <andk@cpan.org>
  • At the beginning of April analysis had to be taken down because of the necessary construction works on Cpantesters 2.0. Since CT 2.0 seems to be coming along very nicely I'm trying to get analysis running again today. It will take a few days before I can say if it is working again. Please be patient, I'll make an announcement on the cpan-testers-discuss mailinglist as soon as I'm confident that we can stay up for a while.

2010-01-31 Andreas J. Koenig <andk@cpan.org>
  • Robert Spier informs me that the outage in LA has an impact on the nntp feed I'm using. http://log.perl.org/ has more about it. I'm seeing grossly dubious calculations on distros that were released before 2009-12-12 but had the third fail after ca. 2010-01-27.

2010-01-18 Andreas J. Koenig <andk@cpan.org>
  • Barbie announces on cpantesters-discuss that the cpantesters master database is marking apocal's reports as invalid.

    Consequence is that at the moment analysis and cpantesters are slightly out of sync because we're not tracking each other's blockings. But this is only a temporary state.

2010-01-17 Andreas J. Koenig <andk@cpan.org>
  • analysis starts filtering all FAIL reports from apocal out because his smoker has gone wild. Currently you may see different counts of FAILs on the home page than on the individual regression pages due to this filtering.

2010-01-01 Andreas J. Koenig <andk@cpan.org>

This site is gratefully hosted on a Dedicated Server, sponsored by www.webfusion.co.uk.