Re: variables in yyparse

[Laurence Finston]

On Sun, 19 Dec 2004, Paul Eggert wrote:

>
> That depends on what you've read.  :-)
>

I've referred to the relevant section in W. Richard Stevens' _Advanced
Programming in the UNIX Environment_, i.e., I haven't read the whole
section.  I know this book must be a bit out-of-date by now, but I still
find it excellent.

I am in the middle of John Levine's _Linkers and Loaders_.  I'm reading it
in preparation for starting to use Libtool for GNU 3DLDF.  Actually, I was
hoping to have an army of developers by now so that I could delegate this
task, but that hasn't happened yet.  I've gotten a bit caught up in other
topics, so I haven't been reading it lately.

> glibc uses the Lea-Gloger allocator <http://www.malloc.de/en/>, which
> as far as I know is documented only in the source code, though there
> are some out-of-date design notes written by Doug Lea
> <http://gee.cs.oswego.edu/dl/html/malloc.html>.

Thanks for the references.  I'll look them up when I get a chance.

> > Is the upshot that `bad_alloc' will never be thrown?
>
> Sorry, I don't quite follow C++ and don't really know the difference
> between the throwing forms of "new" and the others; but the bottom
> line is that yes, if you keep allocating objects and never free them,
> eventually you'll run out of memory (even on GNU systems :-).
>

I've given the matter some thought.  I'm about to check what happens on
the system I'm using when I run out of heap memory.  I hope that
`bad_alloc' is thrown, because segment violations also occur under other
circumstances, so `bad_alloc' would be more informative.

Thanks,

Laurence