Hi,
I've just upgraded monotone's internal botan to version 1.7.3. Some
entropy source files have changed since 1.7.2, thus after minor
adjustments, my debian working box compiles and runs check just fine.
However, I'm not quite sure if Windows boxen still have a botan good
entropy source. Please somebody with access to a windows machine check,
preferably with cygwin as well as native (we support both, don't we?).
I've committed to net.venge.monotone.botan:
7953b7f70906bcf85174245d580df6692aa02d8e [EMAIL PROTECTED]
2008-02-17T14:10:55 net.venge.monotone.botan
As soon as monotone.ca is back up online, I'll push my changes there.
Until then, you are free to pull them from my own server:
nabagan.bluegap.ch (note that only reading is permitted to anonymous users).
You can also find the botan staging branch there
(au.asn.ucc.matt.botan.monotone-2, it allows direct propagation from
botan, thus carries all of botans history with it). As already mentioned
last time I've updated botan, I'm unable to push those revisions
anywhere usable. IMO it would make most sense, to keep those around on
monotone.ca as well.
Doing that upgrade, I'm thinking again about the effort and benefits of
the integrated botan. How expensive would dynamic linking to the botan
library be? The trade off here is: maintaining our own set of changes
and configuration scripts vs maintaining compatibility to multiple botan
versions.
To make use of assembly optimized SHA1, we'd either have to add a
--with-system-botan, and rely on a hopefully optimized system library.
Or add our own architecture detection and conditionally add the most
fitting SHA1 assembler routines. I'm not sure what's easier to do. Any
strong opinions for either variant?
Regards
Markus
_______________________________________________
Monotone-devel mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/monotone-devel
