Re: Serving untracked pages using gitit

[John MacFarlane]

+++ Jonas Juselius [Apr 05 13 10:01 ]:
> I work at a supercomputing center where we manage a large number of
> scientific applications using a nice tool from TACC called Lmod. This
> tool keeps track of all available software modules and related
> information (versions, help text, keywords, etc.). Lmod can generate
> json output of the module data, which we convert to html -> rst ->
> gitit rst pages using a small tool (using pandoc of course!). The
> problem is that, since this is automatically generated data, we don't
> want to put it under git control (the lmod module files are under git
> already). How can I make gitit serve these untracked .page files like
> they were normal pages?

I can't think of any way to do it. Could you perhaps use a
PreParseTransform plugin to transform the raw lmod files to
the desired form? (See docs for Network.Gitit.Interface)
Then you wouldn't have to store a separate converted version.

[Jonas Juselius]

Thanks!

I can't use the raw Lmod files, as they are actually code snippets written in Lua. I poked around in the gitit code, and one solution could be to add a dummy, flat file, amnesiac backend to FileStore (i.e. without revision contol or history). What do you think? I have not looked carefully at FileStore yet, but I suspect it would not be very difficult. Would you accept patches if I wrote such a backend?

Have a nice weekend, Jonsa

[Jonas Juselius]

On the other hand, I can just set up an extra git repository for the generated data. Then the history of the main repository isn't cluttered with automatic commits and data that should not be touched.

j.