On Fri, 4 May 2012 15:04:17 +0200 Joerg Sonnenberger <jo...@britannica.bec.de> wrote:
> On Fri, May 04, 2012 at 10:17:29AM +0900, Cedric BAIL wrote: > > On Thu, May 3, 2012 at 8:57 PM, Vincent Torri > > <vincent.to...@gmail.com> wrote: > > > is it normal that no check is done when calloc is called ? > > > > It's a good question. In E17, there isn't much handling of this > > situation. I am wrong in assuming that on all unix supported by E17, > > when it start returning NULL, it would be way to late ? > > > > Should we handle it ? > > You should check for it and fail deterministically at least. Doing > something like fprintf(stderr, "oops\n"), exit(1); is perfectly valid. > When malloc and friends fail depends somewhat on the setup. Memory > overcommit makes the returned error unreliable in the sense that you > can later get killed by OOM. Using resource limits (ulimit -d / -v > etc) can result in malloc errors much earlier, but whether you can do > something sane from that depends on the circumstances. E.g. if the > user wants to open an image and you can't allocate a buffer, it makes > sense to report an error for that as the image and therefore the > buffer can be huge. Exactly. The response to any sort of memory allocating problem should depend on what you are trying to allocate, and why. Failing to llocate some huge block of memory for something that's not important? Probably not the end of the world, just fail gracefully on the unimportant thing. Failing to allocate a tiny amount for somethung that is very important? The system has probably already spent the last five minutes being unresponsive and thrashing to swap. Might be a good idea to just stop now, coz the OOM might kill us anyway. OR, might be a good idea to free up some large unimportant resources to help with available memory. Just ignoring the failed allocation and trying to use a NULL pointer will likely crash you anyway, but that's just being lazy. Failing gracefully is generally better than failing disgracefully. -- A big old stinking pile of genius that no one wants coz there are too many silver coated monkeys in the world.
signature.asc
Description: PGP signature
------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
