← Back to team overview

bzr-mac team mailing list archive

Re: GUIs for OS X

 

The problem with all the qbzr stuff is how bloody annoying and awkward it is to get going. You have to download QT which is a 132MB download. Then it seems SIP which you need to install from the command line, then PyQT (another 6.6MB) which is another command line install (which I haven't yet been able to install due to errors whenever I try). So that's about 140MB over 3 downloads and 2 command line installs, and all that's just to get the application to run. That is going to put most Mac users off. A native client on the other hand could be < 5-10MB download, drag & drop install and basically "Just Works"™.

Now I haven't been able to try Explorer because I can't get PyQT working. However, if it is anything like almost every other cross platform app then it is going to need a lot of work to make native. Being native isn't just looks but most importantly in function. It's easy to make a text field look like a Cocoa text field for example, but it's hard to implement all the emacs keybindings that all Cocoa text fields support, as well as things such as spell checking. I know QT has been working on a version that wraps around Cocoa, but whether you get that free just by running on the Mac I don't know.

As for the consistency across OS's, I think that depends on the user. I'm writing BazaarX because I want a native client that conforms to all Mac conventions, all the way from install to use. It will take a while to get up to the level of something like Explorer, but it does benefit from using a native toolkit and only having to support 1 platform which can speed up development. That said it is currently a background project that I work on from when I have spare time from the apps that bring in the money.

Martin

On 29 Sep 2009, at 2:11 pm, John Arbash Meinel wrote:

I honestly think that long-term having a OS specific GUI for Mac is a
great idea. I don't think anyone writing that will be able to
reimplement all of qannotate, qlog, qbrowse, etc, in a reasonable amount
of time versus leveraging all the work that is already out there.

It is possible that it is just a 'glue' that is needed, plus a bit of
work to refactor the dialogs to be glued together more cleanly for Mac.
(A dialog may be making assumptions about its menu that should be
changed, etc.)

Anyway, I'm happy to support people who want to make BazaarX, etc
better. I think that with limited resources, focusing on getting Bazaar
Explorer to be better on Mac is a better use of time.

For *me*, I switched to using Firefox & Thunderbird precisely because
when I switch from Windows to Mac to Linux it still works and looks like
what I'm used to. And having consistency *across* OS's was far better
than having consistency *with* the OS. I fully understand the other
side, though, where you are immersed in a single OS and what local apps
to all work together.


References