Jo Dillon wrote:

> > the_buffer
> >       {
> >               data1 {<8><8><8>}          // 3 characters/glyphs/whatever
> >               data2 {<32>}               // 1 character
> >               data3 {<16><16><16>}       // 3 characters
> >               data4 {<8><8><8><8>}       // 4 characters
> >       }
> > the_types
> >       {
> >               type_of_data1 = 8;
> >               type_of_data2 = 32;
> >               type_of_data3 = 16;
> >               type_of_data4 = 8;
> >       }
> >

<snip>

>   This is complicated to manage when you don't know in advance what
> strings you intend to manipulate, and it makes indexing slower than if
> at any one time all your characters are stored in a single array with
> constant-sized members. Oh, and it also increases memory fragmentation.

I understand the indexing part.

But only for my satisfaction: why is memory fragmentation bad? (I'd
understand it if it was the HDD...) I always thought it was good. By
splitting a big chunk of data into smaller pieces uses the memory more
efficiently, because you don't have to alloc a large window, which might
be hard to find in a very solicitated system system. (Or does this only
apply on MS-DOS?)

regards,
florin.

P.S.: Sorry for having sent two messaged with the same contents. I've
made some bad experience with mailing lists, and as I couldn't see my
first mail after more than 10 minutes, I thought I couldn't post to the
list from my home.