Le 27 juin 07 à 13:21, Abdelrazak Younes a écrit :
Mael Hilléreau wrote:
Great! BTW, do you have an opinion (better, an explanation) on
whether it is needed to perform a CRC for graphics? I'm referring
to a previous discussion entitled "Converter problem with Mac OS
packages" (http://marc.info/?t=118019789300008&r=1&w=4). Besides,
nobody answered to my last message on this topic: http://
marc.info/?l=lyx-devel&m=118138743205598&w=4
I don't have a strong opinion on that. I suppose we should first
try the timestamp based solution and evolves to a CRC based
solution if ever needed (and my guess is never).
Could you please recap the problem, the above thread is way too
long for my taste ;-)
;) Well, ... I would say:
A. The goal: we would like to support MacOS packages (which are
actually directories) as graphic files (e.g. Omnigraffle uses such
packages). Despite this is currently a MacOS issue, JMarc said that
it could be useful to accept directories as graphics for other OSes
as well.
B. I proposed a patch which does the job, but in which cache update
is only led by comparing directories' timestamps.
C. Andre and JMarc think that it would be better to use a timestamp
+CRC solution, and not me. I argue that a CRC isn't relevant for
graphics. My message of June, the 9th (http://marc.info/?l=lyx-
devel&m=118138743205598&w=4) recaps those different possibilities and
explains my ideas.
Now, there are two issues:
1. Regarding B.: before adopting this patch, should the code be
enclosed into #ifdefs? If yes, what kind of #ifdefs: IS_MACOS,
HAVE_GRAPHICS_DIR... and where to introduce them? Note also that the
patch preserves CRC+timestamp for regular files.
2. Regarding C.: there was certainly a motivation for introducing the
CRC for regular files, but nobody explained it deeply. So who did
introduce the CRC test, and why?
Hopefully, my way wasn't too long for your taste ;-)
Mael.