[haiku-development] Re: [haiku] Tracker add-on to handle cvs/svn/hg/git?
- From: Ryan Leavengood <leavengood@xxxxxxxxx>
- To: haiku-development@xxxxxxxxxxxxx
- Date: Fri, 14 May 2010 15:22:23 -0400
On Fri, May 14, 2010 at 1:48 PM, Rene Gollent <anevilyak@xxxxxxxxx> wrote:
>
> Right now, there's more or less no mechanism for hooking in in any way
> other than filtering what mimetype(s) your add-on is shown for in the
> add-ons menu, Tracker's add-ons have never been able to make changes
> to the parent menus, icons or anything else unfortunately (yet another
> reason for a refactor/redesign).
I read Axel's old OpenTracker email about his re-design ideas, and I
know Alexandre Deckner has been doing a lot with Tracker lately, but I
suppose I should ask: is it worth tackling at least some of the
Tracker refactor/redesign now, or is this something that must
absolutely be done after R1?
Obviously we should avoid as much as possible adding features and big
changes for R1, otherwise it will never be released. But there also is
pretty much no evolution that can be made in Tracker without this
redesign. Axel seemed to come to that conclusion in 2002, and
Alexandre as well now.
Most people expect things like thumbnails for images, and some ability
to modify/plug-into a file explorer on modern operating systems. Plus
the whole icon overlay situation, which could be useful for the source
control add-on in this thread, as well as for email, bookmarks, and
especially downloading files (which was a great feature from
BeOS/NetPositive which Haiku/WebPositive currently lacks.)
If it helps, I've been thinking about a way to test Haiku applications
using a scripting language, most likely Ruby since that is my language
of choice. But theoretically if we build a way to tap into
applications for testing, then the tests could probably be written in
any language. I want this to build tests around ShowImage so I could
more confidently refactor it, but obviously such a system could be
used to build an automated test suite for Haiku, which would help
immensely for QA and for catching obvious regressions. Plus it can
make developers more confident to tackle refactoring of things like
Tracker as well (assuming a good test suite was written first.)
Of course I have not yet begun on this project, nor do I know how
difficult it might be (though I do at least think it is possible.) Of
course given I have been MIA for quite some time and I don't know what
my schedule may be in the future, there is no telling if I can
implement this in any reasonable time frame. But I will do what I can.
But back to the topic at hand: what is the general opinion on
refactoring Tracker now?
--
Regards,
Ryan
Other related posts:
