Dashcode keeps source files in a “.dcproj” project file, which is really a package to OS X, which in turn is actually nothing more than a directory structure. I’d like to continue to use Dashcode but without the “convenience” of the “.dcproj” structure. Why? I want to…
- check the source code into Mercurial
- use TextMate when developing the non-GUI parts of the code
- eventually build and deploy platform-specific customizations of common core code (i.e., have multiple projects share some common source
I can deploy the Dashcode project into a directory and then do all of the above. But once I do, I likely cannot return to Dashcode, which is quite handy for visual stuff. A more flexible solution would enable me to tell Dashcode where the code is and to read/write it there, not in a “.dcproj” package.
I haven’t heard of anyone doing this – but it would cool because checking a dashcode project into source control is a total pain in the neck (I’m Versions for SVN and it totally chokes on dashcode – I don’t think its Versions’ fault – but who knows..)
I’m finding dashcode quite good for GUI stuff, but its just a neglected product, so there are lots of things that are sort of developer hostile in it.
So here is how I’m using dashcode now:
I use dashcode for GUI layout.
I publish to www_cache_dashcode
I have my custom javascript files in www_client
I then have a little make file that copies everything from www_cache_dashcode to www_client
www_client is under version control, but I exclude/ignore the files that are copied in from www_cache_dashcode
www_cache_dashcode is not under version control
my dashcode project is under version control, but I only rarely update it because there is ALWAYS a conflict. (I’m sure its somehow my fault – but whatever).
Hope this helps,
JJ