Hi Fabio / JrCs
I've finally managed to put together an alternative method for managing the package installer resources. I've updated my branch to work this new way and although it's still a work in progress and not thoroughly tested I would like to communicate what I am trying to do.
The idea is this:
1) Have a single common document which contains all the text for all the files. This file is a Google doc spreadsheet and it's a published file so the data can be downloaded.
2) The chameleon svn will no longer host the individual resources folders and files, instead will host only templates for these files - same as JrCs did for the scripts.
3) The buildpkg.sh script calls a perl script 'ConvertResourcesTextFile.pl' which downloads the data from the Google spreadsheet and then processes it in to a script friendly list format (sort of YAML).
4) The buildpkg.sh script then continues by reading the new list file and populates the templates - building the Resources folder to the correct location for the final package.
You can view a readonly version of the Google doc spreadsheet here:
https://docs.google.com/spreadsheet/ccc?key=0Aj0jJ2rdmK_sdFdNbm45NlpNYU1PcjRmOHVXX0FNa3cI'm still updating the spreadsheet at the moment so you'll notice it's not yet complete. I've still got to add the LocalizationStrings for the Italian language and onwards. Plus make all the recent changes, additions that you Fabio did yesterday.. Lol.
@Fabio - If you would like to help me complete/update the spreadsheet then I could give you write access?
Which brings me on to the question of how to manage the spreadsheet - I don't know if I should make it a public doc so anybody can update it? or only let registered users update it? The plus side of opening it up to everyone would mean it would be up to date quicker and kept accurate, but doing that would mean we also have to trust that nobody messed with it.
This idea is not perfect and I've discovered that the textutil command used in the buildpkg.sh script for converting the unicode markup gives different results in Snow Leopard to what it does in Lion. I need to find out why, but what that means is for now, building the package in Snow Leopard gives the wrong text conversion results.
So for now, that's what I'm working on. Is this a good direction to go in? Have you any suggestions of how this could be improved or changed?
Regards
blackosx