From: Bryon Daly <[EMAIL PROTECTED]> On 2/10/06, The Fool wrote:
> > From: Bryon Daly <[EMAIL PROTECTED]> > > On 2/9/06, The Fool wrote: > >Ok so this is what I've been doing: > > I'll cut down to the meat of it: > > > *((int *) &Mem[PutAt * __Int_Size__]) = PutMe; > > This looks reasonable. I presume that Mem[] is a char or byte array of > some > kind... Are you only writing ints into Mem[]? If you're going to mix > data > types in the same buffer, you'll need to consider data packing or padding, > and possibly be aware of data alignment issues. > > ------- > > The Data I'm Copying from isn't exactly contigous: I meant where you are writing the data to, not from. It looks from your code that you could potentially do something like: PutInt64(1,32); PutInt(2,567); // This int overwrites the upper 4 bytes of the int64 you just wrote or PutChar(1, 32); PutInt(2,567); // Maybe you don't care, but this leaves a 3-byte gap between the 1st char and the int ---- That's fine. I'm probably not going to mix data sizes anyway. It's primarily a place to store data referenced earlier that could be changed back to it's pristine unadulterated state. ---- I also mentioned you should be aware of data alignment: IIRC, bad data alignment only causes a performance hit on Intel, but you'll get a bus error on most other platforms. Your code looks like it will keep things aligned, though, so you should be OK there. Alberto also had a good point about the CurrentSize check. Also note that for PutAt==0, any CurrentSize value >= 0 will get through the check. ---- Hmm. if (((PutAt * __Int_Size__) <= CurrentSize) And ((CurrentSize - __Int_Size__) >= 0)) _______________________________________________ http://www.mccmedia.com/mailman/listinfo/brin-l
