Re: [Comp-hist-devel] Re: Troubles with the C parser

[Hans B P. <ha...@di...>]

Kurt Raschke wrote:
> 
> Well, when I try to run all of comp-hist (in other words the output of 'make dump') through the html parser, look what happens:
> 
>  Date: TBD       ^
> ../../dump[560] Year out of range

Well, that date is not correct!

>  Date: 1992-8-34                ^
> ../../dump[1736] Day out of range

Yes, day number must be < 31

 ...  snip snip ...  more of the same



>  Type: announcemnet       ^
> ../../dump[6110] Invalid field value - using default

Missspelt 


>  Type: Os       ^
> ../../dump[6360] Invalid field value - using default

Accespt os or OS only

>  Name: Apple Computer, Inc.
>       ^
> ../../dump[9145] Duplicate field

Presumably there is already a Name: defined for this node


> Node fortran66, required field Name: not specified

No name specified for fortran66

> be : Cannot find Code taken from  node 'apple'
> next : Cannot find Code taken from  node 'apple'

The node apple is not found

> Node rt11, required field Name: not specified

rt11 has no name node

  ... snip snip ..

> 2.8bsd : Cannot find Successor to  node '2.79bsd 1000'
> 2.79bsd : Cannot find Successor to  node '2bsd 1000'
> 2bsd : Cannot find Successor to  node '1bsd 1000'

Most of the above are nodeids which cannot be found, is it a spacing 
issue? e.g should the last one read "lbsd1000"

> Segmentation fault (core dumped)

Hmm, this one worries me!


Well mostly, the parser is doing its job of validating the input. There
are a lot of invalid datae fields, some Type fields with unsupported
values and node links which don't!

What do you suggest? I can relax the parser checking, but I think the
better solution is to fix the bad fields. I am more than open to
adjusting the parser checks in appropriate ways, eg making the type
field case insensitive would fix one or two issues.

Let me know.

-- Hans