[GRASS-dev] Change in parser behavior?

[Hamish]

Michael wrote:
> It *looks* like changing all options to lower case fixed this. It is  
> NOT an r.terraflow issue because the options are not getting parsed  
> before they ever get to r.terraflow.

I meant irregardless of your script, 'r.terraflow STREAM_DIR=' has upper
case in it so could also be affected by the bug. (see MarkusM's bug report
for that)

Actually now that I look at the code a bit closer technically it's
the last char before the '=' which is tested, not the first. So Option= 
would have been ok but optioN= would not.


> Do you know when this happened? Your email posting was in April of 
> this year.

introduced to trunk and backported to 6.x on 7/24/08:
  http://trac.osgeo.org/grass/changeset/32263

fixed in 6.5svn today:
  http://trac.osgeo.org/grass/changeset/37295

awaiting comments before merging it to other branches.


> It is important to find out when this stopped working for a paper
> Isaac, Helena, and I are about to send in.

6.3.0 does not contain it, 6.4.0rc1 does.

> BTW, all the grass trac addresses (ttp://trac.osgeo.org/grass and the  
> like) produce an Internal Server Error. Is the GRASS trac site out?

yeah I see those 500 errors all the time. Usually followed a few minutes
later by "Database is locked  Oops…  Trac detected an internal error:",
and a few minutes later it works again.

My unsubstantiated theory is that either the machine reboots a lot; the
server software crashes & reloads a lot; or the nightly maintenance scripts
take things down for a little while and I happen to be in a bad time zone
for when that happens. Due to the up&down nature of it my guess has been
more towards the crashing theories.
??


Hamish