The logical place to host this web app would be on the linpvr web site.
Here are a few things that I was thinking of:
1. Configuring the transfer and activation of lircd.conf and lircrc files and port configuration
2. Configuring the creation of the FSTAB file for NFS shares and mount points
3. Configuring Samba shares and mountpoints.
4. Possibly generating server config references for configuring NFS or Samba shares.
5. Generation of localtime files for time zones (if possible).
6. Look at storing settings in mysql, and ditching preapp.scripts?
7. Other suggestions?
A few issues:
1. Is there a need for this? How many people run preapp.scripts, or just don't touch it because Linux is "not their thing"?
2. Of those running custom preapp.scripts, what other things do you use them for? This may also spawn some changes to the cramFS image...
3. What languages does the web server support? I can do PHP, I have more experience under perl, but no real problem.
4. How does the MythTV upgrades alter the database, and is it safe to muck about in the MythTV database or should we spawn our own?
5. If the next major release is NFS based, how would that alter this idea, would it make it more or less important? What changes in functionality would be required?
(please fill out the poll)
I use preapp.script for mounting NFS volumes and configuring LIRC only. I think a web based tool to generate all the required config files would be very useful. I'm no Linux guru, and it still only took me an hour or two to get (a fairly vanilla) Minimyth up and running. The NFS/LIRC learning curve took considerably longer, so a tool that gives you the config files and tells you where to put them would be a great help in initial configuration.
Perhaps longer term, moving towards keeping Minimyth's configurable parameters in SQL sounds like a good idea, especially if it's combined with a web front end for configuring each Minimyth box. I would steer clear of using the myth SQL database as (a) Myth upgrades often change the database schema, and (b) Minimyth development would be tied to Myth development to some extent, as you would not be able to upgrade one without the other (assuming a schema change is involved). On the other hand, I can't really see any significant downside to using a separate database.
Overall, I think the web-based config generator is probably the best way to go in terms of benefit gained vs. development time. This may be particularly true with Minimyth heading towards being NFS based, since the config files could be generated and dropped into palce, and off you go.