skeezix
Internal Development
We fiddled with putting a lame little index on the appended stuff, but never came to fruit; ie: a table of contents. As is, the appended stuff is sort of cheesed .. adding a PXML, and then an icon. If you add somrething else, its difficult to know where the one ends and the other begins, without actually parsing that stuff and slowing things down a bit. PXML ends with a well defined /PXML> and then everything after is icon. But if we start saying .. is this a jpeg? a png? how does it end.. lets find the end, and if theres more, it must be ... a mimetype? an audio sample?
Theres a limit I think to how much you can pack on there. (and doing half-way changes to the pnd format now probably would just confuse people..)
The one idea we had years ago was to append (or pre-pend!) a table of contents block; I think we'd toyed with putting it first in the pnd, and thus the iso was at an offset (which mount can handle just fine), and then using the TOC to jump to the payload items. But it was clunky, and we didn't do that, so.. yeah. (we didn't have a lot of time, we had to design and move fast, back then, for various reasons.. otherwise we could've done a mroe robust solution.)
When we go to .next, we could always try some ambitious new .pnd format (and also keep traditional pnd compatibility of course..)
So, sticking it into the PXML seems a valid solution; making a standard inject/remove tool for devs to use is also valid, but probabyl would get abused or screwed up, so standardizing behaviour in libpnd using PXML seems the best overall .. and not too hard.
*shrug*
Theres a limit I think to how much you can pack on there. (and doing half-way changes to the pnd format now probably would just confuse people..)
The one idea we had years ago was to append (or pre-pend!) a table of contents block; I think we'd toyed with putting it first in the pnd, and thus the iso was at an offset (which mount can handle just fine), and then using the TOC to jump to the payload items. But it was clunky, and we didn't do that, so.. yeah. (we didn't have a lot of time, we had to design and move fast, back then, for various reasons.. otherwise we could've done a mroe robust solution.)
When we go to .next, we could always try some ambitious new .pnd format (and also keep traditional pnd compatibility of course..)
So, sticking it into the PXML seems a valid solution; making a standard inject/remove tool for devs to use is also valid, but probabyl would get abused or screwed up, so standardizing behaviour in libpnd using PXML seems the best overall .. and not too hard.
*shrug*