Re: [gaim-bnet-devel] gaim_normalize()... really needed?

[Gary K. <gr...@re...>]

On Thu, 2005-01-13 at 13:16 +0000, Zilo wrote:
> On 01/13/05 13:50:51, Gary Kramlich wrote:
> > On Thu, 2005-01-13 at 12:21 +0000, evew wrote:
> > > I found a bug inside bnet_buddy_add() in a wrong use of
> > gaim_normalize
> > > ():
> > >
> > > name =3D gaim_normalize(conn->account, buddy->name);
> > >
> > > Using name for saving our buddy inside the hash table is wrong
> > 'cause
> > > gaim_normalize() returns a pointer to a own static array... then =20
> > its
> >=20
> > > value isn't the same during time :P
> > >
> > > As a result of this when receiving friends list all buddy are
> > > considered new (because not found in conn->buddies) then (re)added
> > in
> > > the gaim list... resulting in a big increasing of our popolarity in
> > a
> > > few seconds :)
> > >
> > > Anyway, I was thinking about the needing of this line... 'cause if
> > it's
> > > useful, then we can simply g_strdup() name, but we are already =20
> > using
> >=20
> > > gaim_utf8_strcasecmp() to compare nicks in the hash table... isn't
> > > enough?
> > >
> > > cya
> >=20
> > I thought we were strcmp'n it which would make that legit.. But if
> > we're just comparing pointers, than yeah thats going to be a =20
> > problem...
>=20
> Uhm... I'm not sure I understand your answer... excuse my english, just =20
> in case :P
>=20
> Anyway, we were strcmp'n it... the problem is that the pointer returned =20
> by gaim_normalize() is always the same, and its contents change every =20
> time we call the function.
>=20
> I'm not very familiar with gaim code, so I was asking what is the =20
> meaning of using gaim_normalize()... because if we need it only to
> strdown our names, well, there's already gaim_utf8_strcasecmp()...
>=20
> Well, in others words, I don't know what g_utf8_normalize() is supposed =20
> to do :)
>=20
> cya

As per the documentation at
http://developer.gnome.org/doc/API/2.0/glib/glib-Unicode-Manipulation.html#=
g-utf8-normalize

Converts a string into canonical form, standardizing such issues as
whether a character with an accent is represented as a base character
and combining accent or as a single precomposed character. You should
generally call g_utf8_normalize() before comparing two Unicode strings.

So basically, what I tried to do, but it obviously isn't working, is to
store a comparable string in our hash table, so we can test it before
adding a new one, and therefore avoid duplicates.

--=20
Gary Kramlich <gr...@re...>