The next version of fossil is going to have unversioned files. This is a feature that I have been waiting for quite some time for. Basically it is similar to the Wiki except that it is not versioned. This means that starting with the next version of fossil I can have JavaDoc in my fossil repoository and I can keep it autogenerated via script.


This will be very awesome indeed. I suppose what I should do is create a doclet that outputs to a deterministic markdown format, so that if the source code changes it produces the same document. One issue though would be links to individual methods.


I do wonder though if I should merge it so all projects are considered as one or create documentation for individual projects. One thing though is that the script that performs this will be a bit complex. I have to generate the JavaDoc tree and then go through the fossil unversioned tree and determine which files have been added or removed. I would need to sort both outputs, then when going through each file perform a checksum on it to see if it differs at all. If it does differ then it changes. At least using markdown, the actual output files will be much simpler to read using even simple less.


I should keep the JavaDoc generated sources in its own directory so that it does not mess with anything else. I suppose I can generate it on a daily basis and sync it so that it is up to date. I would keep the script that does it in the repository so that it is versioned and so if I ever migrate systems and such I do not lose the script.


The unversioned files could also have stuff such as videos about SquirrelJME, screenshots, and other things. Although that may bloat the repository a bit.