On 12.02.2012, 13:18 K. wrote:
I've read a few of the CR threads and onwiki discussion about this and I'm surprised by the negativity expressed towards the vol. developer for whatever reason whom is planning to work on this, While yes this extension does plan to have the same or similar feature set in the end, it is planning on using a different method for the end result (as to my basic and non codey knowledge) and trying not to threaten the current usability or ability to push the extension to the cluster in the mean time.
On Sun, Feb 12, 2012 at 5:58 PM, Patrick Reilly preilly@wikimedia.org wrote:
John DuHart has embarked on a complete rewrite of the current Mobile Frontend extension, and has decided to name it 'MobileFrontend2'. While we would prefer it if we could work cooperatively to resolve the existing open issues in the current MobileFrontend extension and maintain continuity. It's understandable why John would prefer to undertake his complete rewrite.
While I don't deny the need of changes in MF, I believe that rewriting production code from scratch a is never a good idea. I don't want to elaborate on this myself, just invite evryone to read this great article by Joel Spolsky: http://www.joelonsoftware.com/articles/fog0000000069.html As the matter of fact, I'm currently in process of refactoring the extension's HTML transformation code, making it isolated, versatile and testable. Believe me, it's much faster than writing from scratch and as a bonus you also retain the original code's long history of bugfixes and improvements - the bonus many people consider to be much more important than development speed.
I think it's actually better completely out from the current extension for a few reasons,
- MF1 is currently a cluster extension so all the code needs to be
reviewed before deployed
- MF1 is already regularly deployed (close to weekly iirc)
- John is working on having it [MF2] operate in a completely different
method than current [MF1] so it would avoid possible breakage and compatibility issues
I think this is an agrument in favor of doing so in a branch, rather than the other way around.
However, we feel that naming the rewrite 'MobileFrontend2' is problematic as users have already started to confuse it with the current extension.
Whom? It's not like it's really advertised anywhere apart from CR and SVN so it shouldn't be causing that many issues at the current stage.
I think Aaron's argument that IDEs may go bonkers from two classes with the same name is pretty compelling, no? What's the problem with doing it in a branch?