Re: [Gnu-arch-users] cscvs log parsing failure

[Charles Duffy]

On Sun, 2003-09-28 at 12:23, Florian Weimer wrote:
> On Sun, Sep 28, 2003 at 12:13:55PM -0500, Charles Duffy wrote:
> 
> > My thought here is that it's necessary to use a parser generator that
> > can generate backtracking parsers, or perhaps to have some readahead (so
> > we can see if the next line matches "^revision " before deciding whether
> > we accept the separator.
> 
> This has already been implemented, and it doesn't help in this case.

Oh -- I just "got" the situation. Damn, that's sticky.

Beyond trying to determine whether the next revision number given is
potentially valid (ie. if it's an obvious backtrack), I don't have a
whole lot of ideas.

Checking revision number validity seems a pretty expensive 60-80%
solution for what I'd *hope* is an odd corner-case.