Re: [Gedcom-parse-devel] Re: gedcom-parse

[Peter V. <Pet...@ad...>]

Hi Steve,

> Thanks, Peter,
> 
> I have a working version of gedcom-parse now that supports TYPE, 
 > FONE and ROMN subfields in NAME, and also FONE & ROMN in PLAC.
> I think I'll finish adding all of the 5.5.1 features, before I 
 > send diffs.  I'm adding these as optional features, and I added
 > an option to turn them on, so you can process "generic" 5.5.1
 > compliant GEDCOMs.

Seems to be a good idea.  Until now, my philosophy was that when
the file declares itself as GEDCOM 5.5, it should be 5.5 and not
5.5.1.  But it looks like some programs just do that anyway, and
it would be nice to support that, instead of converting them to
user defined tags (with underscores prepended).

> 
> One thing I noticed that I don't understand is a call to 
 > gedcom_set_compat_options(0) in gom_parse_file().  I've commented
 > this out in my working version, since I want to use
 > getcom_set_compat_options() from, e.g. gedcom-sanitize, to enable
> 5.5.1 compatibility.
> Do you know why this is being called?  Doesn't seem to make 
 > sense to me.

This is described in the documentation here:
http://gedcom-parse.sourceforge.net/doc/usage.html#Compatibility_mode

gedcom_set_compat_options() does not enable or disable compatibility
mode, it's gedcom_set_compat_handling() that does that.

gedcom_set_compat_options() just gives some control of what the
compatibility mode is allowed to do.  Currently there is only one
option (COMPAT_ALLOW_OUT_OF_CONTEXT), which is useful if enabled,
but can be fatal (with segfaults) to programs that are not prepared
for it.  Have a look at the documentation.

Note that in the CVS sources, this is now enabled in the GOM,
because I've reworked the context handling a bit.  This is not
in any released version yet.

> 
> I would also like to join genes, though I'm focusing on 
 > gedcom-parse at the moment.
> 
> I did notice one problem with genes.  When I have Chinese names,
 > which are formed like
>    1 NAME /Surname/Givenname
> rather than the western tradition of Givenname/Surname/, genes 
 > seems to be losing the given name, or at least it is not
 > displayed.  Probably the code to parse the name did not anticipate
 > that they might be reversed.
> Please be sure the name order from the gedcom file is preserved 
 > when displaying these names, otherwise Asian notes will be mangled.

This is something that Geert will take care of (the split is now
mostly that I work on the parser and Geert on genes...)

Best regards,
Peter.
-- 
===================================================================
Peter Verthez                                     Software engineer
Email:                             mailto:Pet...@ad...
WWW:                           http://users.skynet.be/Peter.Verthez
PGP key:        on homepage (see above) or http://www.keyserver.net
===================================================================
Semper in faecibus sumus, sole profundum variat.