[rrd-developers] getting rid of some of the LOCALE magic

Tobias Oetiker tobi at oetiker.ch
Wed Sep 8 14:03:46 CEST 2010


Hi Pawe,

oops ... bad one ... this exists in many more places ...

thanks!

cheers
tobi

Today Pawe? 'Reef' Polewicz wrote:

>
> Hi
>
> I think I have found a bug in rrd_graph.c related to the LOCALE magic
> mentioned in this thread. I have patched it and tested it and now it looks
> to be working as it should. Here is the patch:
>
> $ svn diff
> Index: rrd_graph.c
> ===================================================================
> --- rrd_graph.c (wersja 2126)
> +++ rrd_graph.c (kopia robocza)
> @@ -3991,7 +3991,8 @@
>      char *old_locale;
>      rrd_graph_init(&im);
>      /* a dummy surface so that we can measure text sizes for placements */
> -    old_locale = setlocale(LC_NUMERIC, "C");
> +    old_locale = setlocale(LC_NUMERIC, NULL);
> +    setlocale(LC_NUMERIC, "C");
>      rrd_graph_options(argc, argv, &im);
>      if (rrd_test_error()) {
>          rrd_info_free(im.grinfo);
>
>
>
> Why did I change this?
>
> setlocale() manual says here:
> http://www.cplusplus.com/reference/clibrary/clocale/setlocale/
>
> """
> Return Value:
> On success, A pointer to a C string identifying the locale currently set for
> the category. If category is LC_ALL and different parts of the locale are
> set to different values, the string returned gives this information in a
> format which may vary between compiler implementations.
>
> If the function failed to set a new locale, this is not modified and a null
> pointer is returned.
> """
>
> So it returns a pointer to a C string that represents the _new_ locale that
> was just set, not the old one.
>
>
>
>
> A test on a patched version:
>
> > /tmp/RRD.py(173)draw()
> -> rrdtool.graph( *args )
> (Pdb) import locale; locale.getlocale(locale.LC_TIME)
> (None, None)
> (Pdb) n
> --Return--
> > /tmp/RRD.py(173)draw()->None
> -> rrdtool.graph( *args )
> (Pdb) import locale; locale.getlocale(locale.LC_TIME)
> (None, None)
>
> Please note that rrdtool.graph() didn't unexpectedly set the locale.
> Now downgrade to debian stable (lenny):
> apt-get install librrd4=1.3.1-4 python-rrdtool=1.3.1-4
>
> > /tmp/RRD.py(173)draw()
> -> rrdtool.graph( *args )
> (Pdb) import locale; locale.getlocale(locale.LC_TIME)
> (None, None)
> (Pdb) n
> --Return--
> > /tmp/RRD.py(173)draw()->None
> -> rrdtool.graph( *args )
> (Pdb) import locale; locale.getlocale(locale.LC_TIME)
> ('pl_PL', 'UTF8')
> (Pdb)
>
> My code has used the python "time.strptime()" method that (indirectly,
> through "calendar" module) uses locale in order to parse dates. When the
> program gathered enough data, it wrote it to rrd, after which the date
> parser suddenly raised exception: ValueError: time data did not match
> format:  data=16/Aug/2010:00:03:20  fmt=%d/%b/%Y:%H:%M:%S.</search-bot-food>
>

-- 
Tobi Oetiker, OETIKER+PARTNER AG, Aarweg 15 CH-4600 Olten, Switzerland
http://it.oetiker.ch tobi at oetiker.ch ++41 62 775 9902 / sb: -9900



More information about the rrd-developers mailing list