On 03/03/2013 06:07 AM, Pete Goodeve wrote:
Hah, again... (:-)) I admit the 0.25 was thouroughly arbitrary; figuring it was a priority, I chose something a little bit higher than... Source Code, I think it was... but I guess it was a bad choice! I'm hoping someone will show up with a full explanation, sometime.
There's code built into the registrar to guess certain text/* MIME types. It is mostly there for text/plain, since there's no obvious way to write a rule for that. If the buffer is long enough, the result has a priority of 0.25. So your rule should have a priority greater than that, if it shall override the built-in code.
BTW, I'm 99% certain that on BeOS the BeBook included a page documenting the MIME sniffer rules (IIRC as a "special topic"). Our BeBook version seems to be lacking that page.
CU, Ingo