pidgin-devel

[Gaim-devel] Re: [Gaim-commits] CVS: gaim/src proxy.c,1.137,1.138 gtkprefs.c,1.292,1.293 util.h,1.82,1.83 util.c,1.368,1.369 gtkutils.c,1.121,1.122

[Richard L. <rl...@wi...>]

On Fri, 2005-10-28 at 16:12 -0700, Sean Egan wrote:
> +	if (!gaim_running_gnome()) {
> +		vbox = gaim_gtk_make_frame(ret, _("Proxy Server"));
> +		prefs_proxy_frame = gtk_vbox_new(FALSE, 0);
> +		gaim_gtk_prefs_dropdown(vbox, _("Proxy _type:"), GAIM_PREF_STRING,
> +					"/core/proxy/type",
> +					_("No proxy"), "none",
> +					"SOCKS 4", "socks4",
> +					"SOCKS 5", "socks5",
> +					"HTTP", "http",
> +					_("Use Environmental Settings"), "envvar",
> +					NULL);

We ended up showing GNOME in the browser preferences list. Shouldn't we
be doing the same here for consistency?

Richard

[Gaim-devel] Re: [Gaim-commits] CVS: gaim/src proxy.c,1.137,1.138 gtkprefs.c,1.292,1.293 util.h,1.82,1.83 util.c,1.368,1.369 gtkutils.c,1.121,1.122

[Peter L. <gai...@si...>]

Sean Egan wrote:
> Update of /cvsroot/gaim/gaim/src
> In directory sc8-pr-cvs1.sourceforge.net:/tmp/cvs-serv1324/src
> 
> Modified Files:
> 	proxy.c gtkprefs.c util.h util.c gtkutils.c 
> Log Message:
> Use GNOME proxy preferences.
> Works For Me. People are going to go crazy.
> We'll try it out in beta.
> 
> 
<snip>

> +		if (!g_spawn_command_line_sync("gconftool-2 -g /system/http_proxy/use_http_proxy", &tmp,
> +					       NULL, NULL, NULL))

<snip>

 From my experience with g_spawn_command_line_sync in /exec plugin, it's 
evil. You're presuming that gconftool-2 will return nicely and in a 
timely manner.

Re: [Gaim-devel] Re: [Gaim-commits] CVS: gaim/src proxy.c,1.137,1.138 gtkprefs.c,1.292,1.293 util.h,1.82,1.83 util.c,1.368,1.369 gtkutils.c,1.121,1.122

[Sean E. <sea...@gm...>]

On 10/28/05, Peter Lawler <gai...@si...> wrote:
>  From my experience with g_spawn_command_line_sync in /exec plugin, it's
> evil. You're presuming that gconftool-2 will return nicely and in a
> timely manner.

Yes, I am. I've been assured by GNOME developers that gconftool-2 will
return immediately and will never print anything other than the value
to STDOUT. Obviouslly, Gaim has stricter requirements for
"immediately" than a shell script would, but if this actually turns
out being problematic, we can change it to use
g_spawn_async_with_pipes(), but that code will be so much uglier.

-s.

--
Author of Open Source Messaging Application Development: Building and
Extending Gaim: http://gaim.sf.net/book.php

[Gaim-devel] Re: [Gaim-commits] CVS: gaim/src proxy.c,1.137,1.138 gtkprefs.c,1.292,1.293 util.h,1.82,1.83 util.c,1.368,1.369 gtkutils.c,1.121,1.122

[Stu T. <st...@no...>]

On Fri, 2005-10-28 at 16:12 -0700, Sean Egan wrote:
> Log Message:
> Use GNOME proxy preferences.
> Works For Me. People are going to go crazy.

Crazy is probably an understatement.

Why has this stuff gone into core and not just Gtk UI? Anyone who
implements an alternative UI (ncurses, Qt, whatever) is going to have
this forced on them.

It falls back to gaim_global_proxy_get_info() if anything is not set,
but there's no way for a user to configure what this uses.

There is no way to configure a SOCKS4 or SOCKS5 proxy that I can see,
and if the GNOME preferences are set to automatic proxy configuration it
won't work.

It'd be nice if we could have such integration, but I fear the
differences between GNOME's proxy support and Gaim's are too great for
it to be realistic.

Regards,


Stu.

Re: [Gaim-devel] Re: [Gaim-commits] CVS: gaim/src proxy.c,1.137,1.138 gtkprefs.c,1.292,1.293 util.h,1.82,1.83 util.c,1.368,1.369 gtkutils.c,1.121,1.122

[Sean E. <sea...@gm...>]

On 10/29/05, Stu Tomlinson <st...@no...> wrote:
> Why has this stuff gone into core and not just Gtk UI? Anyone who
> implements an alternative UI (ncurses, Qt, whatever) is going to have
> this forced on them.

There's nothing particularly UI-specific to it. It checks
gaim_running_gnome() before depending on it. All the proxy code is in
the core, so adding this to UI would have been kinda hacky---even by
the standards set by this hack.

> It falls back to gaim_global_proxy_get_info() if anything is not set,
> but there's no way for a user to configure what this uses.

Actually, that fallback is only if for some reason
g_spawn_command_line_sync() fails. Since I'm doing
g_find_program_in_path() first, I can't think of any good reason this
would fail; I just figured the global proxy was as good a fallback as
any. This could be modified to detect this failure and show global
proxies in the prefs on failure.

> There is no way to configure a SOCKS4 or SOCKS5 proxy that I can see,
> and if the GNOME preferences are set to automatic proxy configuration it
> won't work.

I found some SOCKS settings in gconf I can toy around with.

It should certainly be noted that even if this fails miserably, it can
still be overridden at the account level. Which is, while
inconvinient, good enough for a beta in my opinion.

> It'd be nice if we could have such integration, but I fear the
> differences between GNOME's proxy support and Gaim's are too great for
> it to be realistic.

I don't use proxies, so I really don't care. Also, the preferences
have been slashed enough that I think I'm over that for a while
(especially since now only huge table only appears if you have a proxy
set). It's mostly that we have GNOME people whining at us: you have to
use GConf because you're duplicating proxy settings! How dare you!

And, yes, they have a point that this integration would be nice, so
I'm willing to give it a try so that we can either say "hey, we're
using GConf settings on GNOME now" and they'll be happy or say "we
tried using GConf; it didn't work. Here's what you need to change if
you want us to use it."

Besides, it's not like our proxy support is a bastion of goodness, anyway :=
-p

-s.