[Geany-Devel] small bug in utils_strtod()

[Colomban Wendling]

Le 01/12/2013 23:45, Matthew Brush a écrit :
> On 13-12-01 01:26 PM, Lex Trotman wrote:
>> Hi,
>>
>> **gint** utils_strtod()??  Sheesh.
>>
>> s/b renamed utils_hexcolor() or similar, thats what it does.
>>
>> As for "end", the original strtod() and friends used it to extract
>> several
>> whitespace separated strings from the one string by just passing "end" to
>> the next call, but if we don't use it and its wrong anyways we might as
>> well drop it when we change the name to utils_hexcolor().
>>
> 
> Agree the name is bad, and also the colour parsing isn't very robust. I
> meant to improve it when adding the named_colors support but never got
> around to it.
> 
> Attached patch uses GDK colour parsing instead of own-rolled thing which
> supports more colour specification formats as well as cleans up some
> related code while remaining compatible with legacy colour notation and
> GTK2 and GTK3 builds.
> 
> If nobody objects, I can commit the patch.

As said on IRC, I don't want the color parsing to change depending on
the GTK version Geany is built against.

Attached is a proposal using pango_color_parse() -- which is what
gdk_color_parse() uses.  It supports the 3 and 4-digit channels, even
though we don't actually use such a precision, so it's a bit of a lie,
but I guess nobody cares anyway.

This only renames utils_strtod(), make it stronger and port calling
code, it doesn't include some of the cleanup you did like removing
rotate_rgb().

BTW, I think it's probably better to write "0x0000ff" directly and add a
small comment stating it's white rather than call a parsing function
here, which looks odd.

Regards,
Colomban