[GRASS-dev] [GRASS GIS] #1628: segfault in r.walk

[Glynn Clements]

GRASS GIS wrote:

>  I have removed the lib/sites dependency of v.vol.rst in 7.x in r60272
>  (trunk) and r60273 (relbr7).

That appears to leave v.in.sites as the only client of lib/sites. And
it only uses the "oldsite" functionality (i.e. reading GRASS 5.x
"sites" maps), as opposed to using the legacy sites API to access
vector maps):

 grass=> SELECT i.symbol, i.object
 grass->  FROM obj_exp e, obj_imp i
 grass->  WHERE e.object LIKE 'lib/sites/%' AND e.symbol = i.symbol ;
        symbol        |                        object                         
 ---------------------+-------------------------------------------------------
  G_site_new_struct   | vector/v.in.sites/OBJ.x86_64-unknown-linux-gnu/main.o
  G_site_free_struct  | vector/v.in.sites/OBJ.x86_64-unknown-linux-gnu/main.o
  G_oldsite_describe  | vector/v.in.sites/OBJ.x86_64-unknown-linux-gnu/main.o
  G_oldsite_get       | vector/v.in.sites/OBJ.x86_64-unknown-linux-gnu/main.o
  G_oldsites_open_old | vector/v.in.sites/OBJ.x86_64-unknown-linux-gnu/main.o
 (5 rows)

Is there any reason to retain lib/sites as a separate library, rather
than simply merging it into v.in.sites? There isn't a ctypes wrapper
for it, so I'm reasonably sure it isn't used elsewhere in GRASS.

[Radim: does QGIS use it?]

-- 
Glynn Clements <glynn at gclements.plus.com>