[GRASS-dev] Re: g.mlist

[Huidae Cho]

On Sat, Jun 28, 2008 at 01:15:47PM -0500, Huidae Cho wrote:
> On Sat, Jun 28, 2008 at 08:31:32AM -0700, Hamish wrote:
> > Hi,
> > 
> > re. "sep= changed to separator= while keeping backward compatibility"
> > 
> > two points-
> > ) I think it should be changed to fs= to match the rest of GRASS.
> >  
> > ) backwards compatibility with grass7 (trunk) is nice to do, but it is not at all guaranteed. ie we can break things as we like. it's the dev-branch6 where backwards compatibility must be strictly followed.
> > 
> 
> Hi,
> 
> Yes, fs= looks better than separator=.  If there are no objections to
> breaking this backward compatibility, I'll change it again to fs=.
> 
> > 
> > cheers,
> > Hamish
> > 
> > ps- glad to see the speedups! Now we only have to rewrite that module in C.. parsing the g.list output will always be a little fragile/slow, and a goal of grass7 is to do without helping unix tools so there are fewer problems on the MS Windows port.
> > 

BTW, does this mean we cannot take advantage of backticks (e.g.,
input=`g.mlist pat=time*`) any more?  Then is the only option regex
support for standard options?  Or is it just g.mlist/g.mremove?

> 
> I agree with you that something needs to be done for grass7.  Ideally,
> g.mlist/g.mremove should be removed by extending g.list/g.remove.
> Probably, grass7 will require one more library by default that handles
> regular expressions (e.g., POSIX regex, pcre).  More generally, we could
> implement regex expansion into standard options (too much?).
> 
> Huidae
>