Well, as I finish off my last year here at Seneca College with my friends some would think that our time with Mozilla might be done.  Personally, I don’t think you can get rid of me that easily.  The more rigid release schedules may slow down but my involvement with the open source world is never done (at least not anymore).

After I see my code get into Tinderbox I think I’ll go lean towards a more interesting hobby of mine which is xen virtualization.  Start dabbling in that again as there are quite a few possibilities to explore like VMGL (OpenGL drivers which allow for native fps on virtualized images.  Currently only at OpenGL1.5) or there still being the lack of a management interface to control VMs.

All-in-all, I’ve had a great time working for Mozilla and I think I’ve found a niche for myself in this vast world of software.

Advertisements

Well another release comes with more waiting time from the review team from tinderbox.  Despite clearly stating in a few FAQs on bugzilla not to contact people directly with email I’ve reached an impass.. Looks like some of my code waiting for review on the Make showbuilds.cgi more informative bug hasn’t been touched in almost a month..

I have newer code that I’m debaiting just checking in at this stage.  I’ve held back mainly because I didn’t want to add futher complexity to my already questionable code review (you never know with Tinderbox really).

On another note I’ve updated/fixed/fine tuned another tinderbox release.  All information, as usual, is available at http://zenit.senecac.on.ca/wiki/index.php/Tinderbox_front_page_improvements for review.  New features include:

  • Smoothed out UI by relocated links to make a more vertically compact layout.
  • Bugfix: build machine log files were only displayed for first build machine in tree.

I also introduce comment support today as well.. Look for it in a few hours.  I just have a few bugs to quash and all will be well.

It took me much longer then expected to even get this working.. Tinderbox commenting and documentation are… sparse… to say the least.  That and my love for Perl is particularly minimalistic which makes things ever so much harder as usual.

Sadly its not yet usable as it only displays one log per tree, and not per build machine as was planned.  I debated refactoring some of the tinderbox code to do what I needed only to find that it was far far far to heavily embedded to consider removing.

Without furter ado, here’s a screenshot:

 

screenshot_loglink.png

Exiting no? Yes you’re right, not really. But you would be surprised at how frustratingly long this took to do! Now it’s time for Round 2.  The current code is available on the Tinderbox front end wiki page.

Well.. After using cls’s demo as a starting point I’ve built up the tinderbox front end to look like my original designs release way back. Except now it’s not using json but is generated using Perl code directly in showbuilds.pl. Here’s the work:

TB_v7

It looks good in elinks too (mind you there is no color though) so I’m hoping with a few more tweaks the changes will be close to done.

Though I’d still like to see a few more things added to the front page while I have it in my possession. For example, I’d like to add a link to log files generated during builds as well. The current setup of the front page also overwrites the existing showbuilds.cgi default. I’m not sure if this is a good idea up front because people may not want or may not be expecting a front page change.. Therefore I’m assuming putting a link from showbuilds.cgi to the new front page may be a better idea until the page is better adopted by the public.

This release also uses some CSS!  Therefore users will be able to craft their own stylesheets to  make tinderbox look as they please…

The fron tend for Tinderbox will now be based off of the first draft that I posted for it.  Ever.  I’m now using cls’s Tinderbox example and just shoving all of the good looks into that little package.  It’s even possible that CSS will make it into this release as long as it passes the tests.

Testing is done by browsing the site using elinks.  If all looks good then it’s possible I’ll be in the clear.  I’m also debating about dusting off my old home server and slapping Tinderbox on it so users can view my work.  The challenge isn’t the install as much as it is the relatively delicate and long install process of Tinderbox.  But it will get done.

After I have Tinderbox available publicly again I’ll have some easy contrib work to do!  Basically consists of elinks + browsing my work to ensure there are no bugs, errors, or the like.

Well, it seems that the over hasty plans for using JSON output on the Tinderbox application may not end up being realized.  The main reasons being the usual assortment: security issues, Tinderbox becoming too complex, speed issues, and demand being low for the feature.

The JSON feature will be dropped in hopes of accomplishing the primary goal of building the new user front end.  Since the HTML will not change that will not need to be rewritten.  However, the AJAX/JS will be dropped to provide a static front end.

Well.. with this release I’ve worked out a lot of kinds in the front end code for tinderbox. I’m hoping that the front end will be informative in its current state and that people will be able to get good use out of it. So without further a due, here’s a quick screen shot.

Tinderbox with color

The amount of trees and build machines is.. lacking.. to say the least. I’m working on a local machine and its quite annoying having to manually send 30 email every few hours to create a realistic working environment.

As soon as the perl code is committed to the real tinderbox project I will be able to use live data. Then it will be grand!

Well, construction of the Tinderbox UI moves forward.. I’m still working out some details but all in all things more forward.

I should be releasing some new contribution ideas shortly so stay tuned for that as well.

Well, after a bit of silence due to the Christmas break I’ll finish rolling out my Tinderbox improvements.  There is not much left to do in terms of the front page aesthetics, however, there are still a few more ideas that I would like to apply.

One idea being the ability to parse Tinderbox information in a more efficient manner.  This could include placing Mozilla build information into tables, graphs, and other more readily readable formats.  As it stands now, gathering and analyzing large amounts of information is very time consuming.  There is no one location that a user could go to gather the information and is forced to check each build individually.  This should all change with the introduction of my concept which will localize information into a singular page or area.

My Tinderbox front end code has been slightly delayed for release because of the joys of UNIX! It’s always a surprise working with UNIX and I guess today was no different. After accidentally hibernating my computer running the regular Ubuntu kernel and rebooting it into the XEN Ubuntu kernel things got a little messy.  I’m surprised there was no safeguard against this as it caused all hibernated programs to immediately segfault and lockup my computer.  Upon restart the fdsk utility promptly repaired about four thousand broken files, links, and other things (which is NEVER good).  But as mentioned, the data was still solid and intact..

After a few nights of trying to learn the jist of Ajax, JSON, and JavaScript I’ve finally got a working version of the Tinderbox front page that is generated dynamically.

To make this all work there were a few modifications that were made to tbglobals.pl.  This allowed me to output a list of Tinderbox trees in JSON which was not previously implemented.  Afterwards, the index.html page has some JavaScript that parses the JSON output as I need it.

All in all, it works!

Tinderbox 0.3a Release

All in all, it was an interesting experience working with Perl and JavaScript again, I’d forgotten the power of them both.. but mostly JavaScript as I’m a PHP guy myself!