Thanks for the quick response. On 3/25/14, 9:19 PM, Allan McRae wrote:
Did this even compile? The include should fail given libalpm is not in the pacman source inlcude path.
It does indeed compile, I'm using it! :) The header I include (handle.h) is in the same directory as the header included immediately above it (alpm.h). I built it with makepkg; are there other environments in which you expect this to fail? It seems to me that libalpm is always distributed with pacman.
Anyway, as with all the other options you will need to add a alpm_option_set/get_...() function. This gives all frontends access.
No documentation updates to be seen. pacman.conf man page and example pacman.conf will need changed.
Indenting is always tabs in pacman. This varies throughout your patch...
I will send a patch shortly addressing these mistakes. Does libalpm have tests that I should also be updating?
How does curl handle negative values for the limits? Perhaps we should ensure this is positive too.
It's a simple less-than comparison; see curl's lib/speedcheck.c:39-57 for details. Setting a negative low speed time will always trigger the low speed failure, e.g.: Operation too slow. Less than 1 bytes/sec transferred the last -30 seconds Setting a negative (or zero) low speed limit will effectively disable the check, regardless of the low speed time. It may be more elegant to provide an option to disable the check, but I usually err on the side of flexibility. Thanks, Andrew