> On Thu, 2008-01-24 at 09:57 +1300, Amos Jeffries wrote: >> > >> > On Wed, 2008-01-23 at 19:07 +0200, Tsantilas Christos wrote: >> >> squid crashes after this patch, something is wrong. After reverting >> to >> >> a previews version of String.cc all are OK again ..... >> >> >> >> Amos Jeffries wrote: >> >> > Reduce call duplication in String. >> > >> > I committed a fix to HEAD. However, I see other suspicious behavior >> when >> > doing ICAP so there may be more problems to fix. >> >> Definately. I'm not finished there. >> This was just one of the validation failures in String. Thanks for >> finding >> the caller. I've also beefed up limitInit itself to handle the simple >> 0-len case it should be able to cope with by itself. > > I did find more problems caused by the "call reduction" changes and will > commit a fix soon. I am going to change a few names and make a few calls > private to avoid this kind of confusion in the future.
FYI, I'm working toward a single limitInit(str,len), clean()/reset(), and constructors. > > Amos, you may feel like you were cursed by SquidString, but the reality > is that the core API problems were introduced or not removed much > earlier, when C strings were half-converted to C++. You are just being > bitten by the code that is very difficult (and, hence, dangerous) to > modify, even a little. Sigh. I'm finding that. Thus the miniscule changes with major input validation tests. Amos
