Re: [Confuse-devel] issues in 2.5

[Martin Hedenfalk]

Hello Dan,

The purpose of the CFGF_NODEFAULT flag is to prevent libConfuse from 
assigning an option a default value. This can then be used after 
parsing to detect that mandatory options have not been specified in a 
config file.

A CFGF_MANDATORY flag has limited usefulness. First, when should that 
flag be checked? It can't be check in cfg_parse, since the caller could 
continue parsing other config files. Second, it is easily implemented 
by the caller by checking cfg_size() of the mandatory options (assuming 
they don't have a default value, hence the CFGF_NODEFAULT flag).

Please reply if you have another opinion, or a suggesting of how to 
implement it.

Thanks for your interest in libConfuse. I'll look into the cfg_print 
bug and update the docs.

Regards
Martin

2005-01-07 kl. 22.26 skrev Dan Lipsitt:

> Here are some things I've noticed about the current version of 
> libconfuse:
>
> 1) The CFGF_NODEFAULT option doesn't behave the way I expected it to.
> I assumed (perhaps incorrectly) that if an option had CFGF_NODEFAULT
> specified and a value for that option was not specified in the config
> file, then an error would occur during parsing. If this was not the
> intended function of CFGF_NODEFAULT, it would be a useful one to have.
> Perhaps another flag could provide it. Something like CFGF_MANDATORY.
>
> 2) If a CFG_STR defaults to "" (an empty string) and isn't set in the
> config file, cfg_print() will often print junk from memory for the
> value of that string. However, I am having trouble producing a short
> program that replicates this problem.
>
> 3) The version of the documentation at
> http://www.nongnu.org/confuse/manual is still at 2.3.
>
> Thanks, Martin for such a useful library!
>
> Dan