Monday, April 22, 2013

Status report: 2013-04-15 - 2013-04-21

A really long time without activity. AsiaBSDCon 2013, vacations, coming back to work, and a toddler with self-mobility kill all chances of having some spare time. Anyway:

  • Implemented support to filter particular result types out of HTML reports. This is a requisite for generating reports for the HTML test suite, as including details for all passed results increases the size of the reports (and the number of files to sync across mirrors) significantly.
  • Fixed some build failures in FreeBSD 9.1 powerpc64 with Clang.

Tuesday, March 12, 2013

Status report: 2013-03-04 - 2013-03-10

  • Modified Anita to transparently use Kyua when the NetBSD system being tested was built with MKKYUA=yes. Sent the patch for review and I'm now awaiting integration upstream.
Other than this, there hasn't been much activity last week. I've been incredibly busy closing things out in preparation for AsiaBSDCon 2013. As a matter of fact, I'm almost on my way to the airport! Unfortunately, I didn't expect to have almost-finished the import of Kyua into NetBSD by now, and that's the reason why I didn't apply for a slot during the Call for Papers :-(

Thursday, February 28, 2013

Initial Kyua import into NetBSD done

I am very happy to announce that the initial import of Kyua into the NetBSD source tree is complete! It has been a really long journey and this is the first step towards completion. No, this is nowhere done yet: while the code is in the tree, there is some more integration work to be done.

Monday, February 25, 2013

Status report: 2013-02-18 - 2013-02-24

There is only one big —well, actually two— news items this last week:
The goal set two and a half years ago has finally materialized, although it's nowhere near completion.

As a quick note, you can start playing with this by setting MKKYUA=yes in your build configuration. I still have to ready some minor changes and send a formal announcement with all the details. Stay tuned!

Friday, February 22, 2013

Kyua 0.6 released

After a really long wait —wow, it has been 8 months since 0.5; way too long— Kyua 0.6 is finally out!

This release brings in tons of changes, and is the release that will go into the NetBSD tree this weekend as soon as I finish doing some local testing of the integration changes. Many of the changes in this release are the result of the user feedback gathered from the proposal to import the code into NetBSD.

For the full list of changes, please see the kyua-cli-0.6 release file page.

The pkgsrc packages have already been updated to this new version and the update in Fedora is blocked by the review of the kyua-testers package. Help in the latter appreciated!

Tuesday, February 19, 2013

Status report: 2013-02-11 - 2013-02-17

  • Released ATF 0.17 and imported it into NetBSD.
  • Added caching support to the configure tests of kyua-cli so that cross-building is possible.
  • Started import of Kyua into NetBSD by adding support for MKKYUA to the build system and importing Lutok 0.2.
  • Build-tested the preliminary release of kyua-testers under NetBSD i386 and amd64, and Fedora 18 and rawhide.
  • Wrote the pkgsrc and RPM packages for kyua-testers.
  • Prepared changes to import kyua-testers into NetBSD.
I'm working on the kyua-testers import into NetBSD as we speak, but this could not happen in time for last week's report (aka this report) ;-P

Monday, February 11, 2013

Status report: 2013-02-04 - 2013-02-10

I spent 10 hours in a bus this week and had some incredibly productive time:

  • Fixed kyua help to not fail when the configuration file is bogus. Getting help is something that should work all the time. (Issue 36.)
  • Simplified the syntax() call in configuration and Kyuafile files to not include the format name. The format name is implied and was unnecessary. (Issue 37.)
  • Added the metadata of test cases to HTML reports. This is handy for postmortem debugging.
  • Stopped wrapping text on screen boundaries for most messages printed by kyua. Doing wrapping with newline characters broke copy/pasting and was incredibly annoying. This now only happens in the output of kyua help, which is the only thing that should look "nice".
  • Rewrote the Kyuafile parser in C++ instead of relying on an external Lua file. The resulting code is lightyears clearer and helped to removal of a bunch of poorly-understood code. The kyua binary is now self-contained (modulo some non-critical distribution documents).
  • Fixed the configure checks that look for GNU Getopt to work in cross-compilation environments. (In ATF only; still getting this to work with the Kyua tests.)
  • Fixed a bunch of run-time issues found when running under Debian testing on my shiny-"new" PowerMac G5.
  • And, lastly, updated the import plan into NetBSD and notified tech-userlevel@ about the impeding changes.
So, stay tuned: I shall start importing a new release of ATF and Lutok into the NetBSD source tree Real Soon (TM) now. Once those are done, kyua-testers and kyua-cli will follow after they pass release tests in a variety of systems. core@ has given permission to do these imports as incremental steps, provided that they respect the new MKKYUA variable, which will simplify this process significantly.