Re: [ANNOUNCE] GIT 1.5.3-rc4
- From
Steven Grimm <koreth@midwinter.com>
- Date
- Aug 4, 2007, 03:49 UTC
- Message-ID
- <46B3F762.1050306@midwinter.com>
- In-Reply-To
- <7vsl70jdcr.fsf@assigned-by-dhcp.cox.net>
Junio C Hamano wrote:
> The asciidoc toolchain used by us (either AsciiDoc 7 nor 8) does > not seem to work well with docbook-xsl 1.72 and 1.73, it seems. >
How attached are we to asciidoc? Every time I do a clean build and sit there twiddling my thumbs waiting for xmlto to do its thing, I think to myself, "If this were a dedicated Perl script to do the syntax transformations directly to man and html formats, it would blast through all the .txt files in a second or two total." It seems outlandish to me that it takes longer to build the (relatively small) documentation than it does to build the actual code. Plus we constantly run into this sort of problem.
Do we want to keep using asciidoc (e.g., so people can easily export to other asciidoc-supported formats), or is a dedicated renderer something we'd consider switching to? I have a flight from China back to the US coming in a couple weeks; this could be a perfect little project to keep me occupied between in-flight movies. It doesn't look like the syntax transformations are very hard, and it'd be easy enough to verify correctness by just comparing against the existing asciidoc output.
Am I correct in observing that "*roff -man" and HTML are the only two output formats we care about, or do people use other formats in their private branches?
-Steve