Re: [Sc2-devel] strcpy to strncpy

[Serge v. d. B. <sv...@st...>]

On 6 Dec 2002, Abaddon wrote:
> ok, heres the thing...these fixes are to effect stability, in a proper
> running environment they should never overrun anyways, and up to this
> point everyone has been just hoping that was the case...these patches
> are just to effect stability...more to the point they are to make the
> propagation of bugs harder...
>
> the original code doesn't even check to see if its going to fit, with
> these patches it will always fit, it might crash a second later if its
> not null terminated but it didn't trash your stack or your heap...the
> problem is that this often doesn't crash the system but causes data
> corruption...
Yeah, I understand this must be annoying to you. And I agree that the sort
of incorrect behaviour caused by cut off strings is usually easier to debug
than that caused by a corrupted heap. However, wasn't the idea of these
patches to get these things fixed once and for all?

If you don't want to fix this, I'll apply your previous patches, as they
ARE better than what we have right now, but I'd be much happier if you'd
take another look at it. It's your choice.

> tell you what, you guys decide how you want this fixed and ill fix it,
I'd like to see too long strings be detected and handled.
How you do it, I don't mind much. The stuff I mentioned in my previous mail
were only suggestions. (just keep in mind it needs to run on various
platforms).

> but when you do finally decide how this is best fixed, please please
> please please, make sure everyone from then on out follows the
> conventions because these patches,
We've got the 'Contributing' file now for precisely this reason.

Serge


Btw, I'm at the moment catching up with some other work I should have done
before, when I was working on UQM. Don't expect me to have much time to do
many commits for a few days.