stop duplicating example-contents data

Bug #612563 reported by Martin Pitt
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
checkbox (Ubuntu)
Fix Released
Medium
Marc Tardif
Maverick
Fix Released
Medium
Marc Tardif

Bug Description

Binary package hint: checkbox

As part of investigating why the CDs are once again oversized since alpha-2, I discovered that checkbox balooned excessively:

checkbox (Δ 5.9 MB - 0.10: 0.1 MB 0.10.1: 6.0 MB)

This is because it started shipping copies of data files which we already have (mostly) in example-contents. This is a major waste, and we should just add a dependency to example-contents instead.

Martin Pitt (pitti)
Changed in checkbox (Ubuntu):
importance: Undecided → Medium
Martin Pitt (pitti)
Changed in checkbox (Ubuntu Maverick):
milestone: none → ubuntu-10.10-beta
David Murphy (schwuk)
Changed in checkbox (Ubuntu Maverick):
assignee: nobody → Jeff Lane (bladernr)
Revision history for this message
Jeff Lane  (bladernr) wrote :

There were some valid reasons for not using example-contents as a dependency for the checkbox tests that require data similar to the ones in question.

The biggest one being, IMHO, the need to not depend on example data that could arbitrarily change, or disappear altogether in the future. Another being the need for file names that make sense from a testing perspective: SongTitle-ArtistName.oga is not a useful description for test output while OggVorbisAudio.oga is.

Also, the structure of the ./data dir in Checkbox was changed to accommodate sorting the various sample file types by category (Audio, Video, Documents, Web) so we could have tests simply parse a directory and tests against the types it finds in there. This allows us to run smaller test sets targeted against specific types or applications (e.g. we can run a batch test against common video types, or common Document types).

One thing I noticed, however, is that the Alpha 2 isos have an version of Checkbox that is not current with trunk. The current trunk has full directory structure and descriptive file types (including types that aren't part of example-content) and was created to satisfy the requirements of the consumer test list project for certification. I'm assuming the current tree has the latest checkbox trunk and will appear in Alpha 3 and beyond.

So for now, the decision made is to move all the data and associated tests from Checkbox to Checkbox-compatibility and that should be accomplished prior to the Beta spin.Bug Description

Changed in checkbox (Ubuntu Maverick):
status: New → In Progress
Revision history for this message
Martin Pitt (pitti) wrote : Re: [Bug 612563] Re: stop duplicating example-contents data

Hello Jeff,

Jeff Lane [2010-08-03 21:05 -0000]:
> Another being the need for file names that make sense from a testing
> perspective: SongTitle-ArtistName.oga is not a useful description for
> test output while OggVorbisAudio.oga is.
>
> Also, the structure of the ./data dir in Checkbox was changed to
> accommodate sorting the various sample file types by category (Audio,
> Video, Documents, Web)

Could the files which are in example-contents merely be symlinked
instead of being full copies? Then you can retain the naming schema
and the ./data structure.

Thanks,

Martin
--
Martin Pitt | http://www.piware.de
Ubuntu Developer (www.ubuntu.com) | Debian Developer (www.debian.org)

Revision history for this message
Jeff Lane  (bladernr) wrote :

After discussion with Martin and the original plan to just move everything, I settled on a compromise.

1: I moved most of the extraneous files to -compat for now.
2: Created a smaller subset of doc/file type tests by using symlinks to some of the items in example-content

This got Checkbox itself back down to a little less than 800K

This, I feel, is a good compromise that gives us at least some basic file type testing in the general public AND allows us to keep the Consumer Tests intact by moving them to -compat.

Changed in checkbox (Ubuntu Maverick):
status: In Progress → Fix Committed
Revision history for this message
Martin Pitt (pitti) wrote :

Thanks Jeff, appreciated!

Revision history for this message
Marc Tardif (cr3) wrote :

The files themselves were refactored to concentrate on what was being tested precisely. In the end, the same feature set was preserved and the overall deb size is now 1.2M.

Changed in checkbox (Ubuntu Maverick):
assignee: Jeff Lane (bladernr) → Marc Tardif (cr3)
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.