On Thu, 7 Feb 2002, Dave Neary wrote: > Marc Lehmann wrote: > > > There is no such thing as "non-standard naming" in this case. exif > > doesn't provide a standard name, so you need to make one up. Wether > > oyu make up one or 50 doesn't really matter, as long as it's > > documented. > > That's the major part of the problem. If someone adds metadata that's > not documented, that's a problem. It's a problem because there's no > one place that someone can go and say that they have the definitive > list of parasites. That's fine if there are only a few, but if there > are many, tracking them becomes a chore. And more it leads to the > possibility of inconsistency. [snip] > > I think it's much easier to just look at the documentation rather > > than to run through header files until I can find what I need, > > hopefully with a sparse one-line comment, even. > > That's fine, when the documentation is accurate - if it lags behind > (and anyone who thinks it doesn't is living in a dreaml world) it > becomes at best redundant and at worst misleading. > > > parasites were created for metadata. If they don't work well > > enough for that parasites should be improved, rather than becoming a > > legacy layer. > > To handle something like what we're talking about, you'd need > something paralleling the PDB - a parasite database. And that seems > to me like overkill. While I think that the problems with inconsistancy are overrated -- mitch and sven seem to review every cvs commit and wouldn't reluctant to revert a commit that adds a parasite without adding it to devel-docs/parasites.txt -- I also don't see why adding a PDB-style self-documenting parasite registry would be bad either. The advantage would be that there would be one place to read about all the parasites, even those that are not included with the GIMP. Before a plug-in could attach a parasite, the parasite would have to be registered. The registration function would take three parameters: the name of the parasite, a blurb about what it does, and the format that the parasite is stored in. (UTF-8, 7-bit ASCII, string representation of a float, etc.) Perhaps even a custom format could be provided that calls a custom verification function. Since more than one plug-in might use a parasite, but all of them have to register the parasite, it's possible that the information registered can conflict. It seems best to make the second registration function fail should the formats be different, but if the blurb is different, it is probably best to silently ignore it. What do you think? Rockwalrus
