Michal Čihař - Archives

Semantic phpmyadmin.net

Today I published set of changes to phpMyAdmin website, which add microformats attributes where applicable. Currently hCard and hAtom were added. The first one enables you to grab contacts directly from web page (if your browser supports it, currently you need some extension such as Operator). I also added some rel attributes to links, so application browsing the website can know relevance of some links.

Gammu test version 1.21.93

After few weeks of debugging build system, new Gammu version is out. This time most fixes are in build system area, especially now it is possible to compile Gammu using Borland C++ 5.5. Also my hacks for crosscompilation to Windows on Linux have been dropped in favour of crosscompilation features in CMake 2.6.

Full list of changes:

  • Drop own crosscompilation support, rely on CMake one.
  • Also standard CMake parameters used when possible (eg. BUILD_SHARED_LIBS).
  • Search does not create thread for non existing devices.
  • Fixed possible crash when backing up ringtones.
  • Fix possible race condition in searching.
  • Fixed invoking of CMake with output to custom directory.
  • Compile on platforms where ssize_t does not exist.
  • Fix detection of towlower on some compilers.
  • Make Bluetooth compile using BCC.
  • Fix compilation if strtoull does not exist.
  • Compile on platforms where intptr_t does not exist.
  • Fixed compilation in BCC (see changes above).
  • Check whether compiler supports __FUNCTION__.

You can download from usual place: http://cihar.com/gammu/,

Debian users will find packages in experimental soon.

So simple patch...

Today again somebody commented on probably simplest patch I ever attempted contribute to some free software. And it memorized me the funny fact that this simple change lies for five years in bugtracker without single comment from developers. Only occasionally some random hacker founds it and ask why such simple issue has not yet been solved. Latest change in bug adds flag, that this is still broken in all recent Python versions including 3.0 :-).

I'm not a robot!

Dear Google,

can you please tell me, why I have to retype f**king captcha, every time I want to use searching? Guess what I'm even logged in using Google account and I've already retyped several captchas, so I would have been really clever robot, if I would be able to do this.

Thanks

PS: I've tried different set of parameters to search, but it does not matter what I type. The only thing which does matter is that I'm connected through local provider which does NAT of hundredths computers to single IP address and some of these computers probably did something evil to Google.

phpMyAdmin has new website

I really did not like phpMyAdmin project website for some time and wanted to change it. However it is harder and harder to find time to do some bigger thing, so this task was stuck in my queue for quite a long time.

One of biggest problem is that the website is hosted on SourceForge.net and they limit quite a lot what you can do there. Especially including external resources is hard - the website should include lot of information from SourceForge.net as news posts, file releases etc. The only reasonable option to get all external data into the website is to generate static content and push it to SourceForge.net web server.

This decision was quite clear to me and it lead me to using Python with Genshi templating to generate the website. Yes, you read it correctly - one of most popular PHP applications is using Python to generate its website.

I started to play with Genshi and feedparser about one month ago, with no clear concept of pages design, just to learn the technologies. However after creating first proposal and showing it to some friends, phpMyAdmin team members and frequent visitors to phpmyadmin IRC channel, I was quite surprised about positive feedback, so I started to work a bit harder to make it possible to have new website before we release 3.1.0.

And here it comes, just check it out at http://www.phpmyadmin.net/.

Gammu test version 1.21.92

Good news everyone, new Gammu testing version is out. This time biggest fix is Bluetooth support for Mac OS X, accompanied with fixed locking on some architectures and improved debug configuration, to allow proper integration with python-gammu.

Full list of changes:

  • Reimplement locking and add tests for it.
  • GSM_SetDebugFileDescriptor now accepts flag whether file descriptor can be closed (bug #749).
  • Soname change due to API breakage (see above).
  • Fixed compilation on Mac OS X, thanks to Juan A. Bertolin for testing.

You can download from usual place: http://cihar.com/gammu/,

Debian users will find packages in experimental soon.

Firefox hidden config of the day

browser.tabs.loadDivertedInBackground = True and Firefox 3 no longer steals focus when I open link from external application.

PS: This seems to be most interesting post in my blog, at least when counting email feedback - looks like there was lot of people annoyed by this behaviour and too lazy to search for solution :-).

Gammu test version 1.21.91

Good news everyone, new Gammu testing version is out. Fixes are there for everybody, starting in AT engine initialization, going through FBUS and Nokia drivers up to OBEX handling in Siemens phones.

Full list of changes:

  • Script gammu-config can now also work with whiptail.
  • Minor fixes in gammu-config script.
  • Extra spaces from connection or device name are now stripped.
  • Automatically reinitialize FBUS connection if it is dropped.
  • Disable OBEX with Motorola Z6c, it is broken (bug #746).
  • Refresh manufacturer information on second init of AT (bug #746).
  • Always get charset information on AT phone initialization.
  • Better handling of OBEX switching for Siemens phones.
  • Fix flags for 2630 and 3109c (bug #752).
  • Fix handling of caller groups and picture IDs (bug #752).
  • Properly set calendar location while adding (bug #747).

You can download from usual place: http://cihar.com/gammu/,

Debian users will find packages in experimental soon.

Bad translations status

I know phpMyAdmin translators will love us for this, but I just commited translatable messages for new setup script. This leads to situation when even the best translations are at 75% (see statistics). I hope our brave translators will manage to handle these 350 new messages for upcoming 3.1 release, at least for major languages, which were always close to 100%.

PS: Looking for translator to Czech!

Double check what you do

I just wanted to fix some RC bugs till I'll go home, so I looked at the list of open ones and randomly chosen two bugs about roxen fonts which looked like low hanging fruits. Both were in fact same issue, just copied to two packages.

I worked on the fix, creating some intermediate version and then I finally wanted to upload the package. Unfortunately in this step I somehow managed to use the intermediate version, which did not really fix the issue. As packages were already uploaded, there was not much I could do except yet another NMU, now with full set of changes.

Just as accepted emails arrived I also noticed that I messed up bug numbers in changelog - I used same bug number in both changelogs. Okay, manually closing the bug worked around this.

But now I'm afraid what else did I mess up. Probably it's time to go for a dinner and stop working on computer for now :-).