I would suggest to exclude the test files (sapphire/selenium, sapphire/simpletest, and maybe also sapphire/tests) of the standard download since most users won't ever use or need those files and it would save some bandwidth and storage and speed up a little bit the manifest building.
This would reduce the size of the download (v2.0.2b) from 2.03 MB to approx. 1.80 and save almost 1 MB after unpacking it.
What do you think about this idea? Nevertheless you could offer a separate archive with those test files.
They're currently not used as part of our build process. We're going to be setting up a suite of tests based on PHPUnit shortly.
Until there are more than a couple of test files, I think that removing them from the distribution is premature optimisation. You'd get a lot more mileage, for instance, in removing the TinyMCE plug-ins / languages that we don't use! :-)
If we're not using something, why should it be in there? Especially if it reduces the file size of the download by a couple of hundred KB. Sam, will we be removing these eventually in favour of PHPUnit or keeping them in there?
Having TinyMCE more modular would be helpful. That massive batch of language support files is irritating to say the least when uploading via FTP. Agreed Will! :p
> You'd get a lot more mileage, for instance, in removing the TinyMCE plug-ins /
> languages that we don't use! :-)
> If we're not using something, why should it be in there?
> Especially if it reduces the file size of the download by a couple of hundred KB.
Well, I think it would be also a good idea to remove the unused TinyMCE plug-ins but since Bernat is working on localization of the whole system in my opinion it would be a step backwards by removing the language files.
Another important aspect is that by removing unused files we not only reduce the file size but also the attack surface!