[Spice-devel] [PATCH spice-gtk v2 0/4]Add more tests for the session's uri

[Marc-André Lureau]

Hi

----- Original Message -----
> Hi,
> 
> On Wed, 2016-10-12 at 11:38 -0400, Marc-André Lureau wrote:
> > Hi
> > 
> > ----- Original Message -----
> > > Hi,
> > > 
> > > I plan to merge both uri parsers (one is in spice-session, one in
> > > spice-uri).
> > > These patches add more to tests to avoid regressions.
> > > 
> > > v2 per Victor review splits IPv6 tests and checks for expected
> > > warnings
> > > v1: https://lists.freedesktop.org/archives/spice-devel/2016-May/02
> > > 9261.html
> > 
> > Reminds me of https://lists.freedesktop.org/archives/spice-devel/201
> > 5-March/019288.html
> > 
> > It's a shame to see glib so slow at picking the GUri patches. I
> > personally lost interest, there are other better ways imho.
> Yes it is
> 
> > 
> > Tbh, instead of slowly reinventing YA URI parsing in C,
> 
> The parser is already implemented in spice-gtk (twice). My intention
> is to have just one parser in spice-gtk, after that it should be
> easier to migrate another parser, fix bugs etc...

Sure, but merging the two will probably result in more complex URI (and yet incomplete or incorrect) parsing.

> 
> >  I am tempted to go a step in the future and just link with the rust
> > URI crate (https://docs.rs/url/1.2.1/url/). Since it's already being
> > used by firefox in the latest release, I trust this is fairly solid.
> 
> Probably it is and rust is interesting, but imho it is overkill to use
> rust just to have better/safer url parsing :)

Well, it would have to start somewhere. That's what firefox did (along with mp4 parsing iirc). I would be really glad to use a safer language with a very active community (contrast to glib development, stable or moribound depending on pov)

> > I have done some mix of rust and C in the past, just like what Hub
> > recently described in a blog: https://www.figuiere.net/hub/blog/?201
> > 6/10/07/862-rust-and-automake
> > 
> > It is fairly easy, and I would be happy to link with rust code which
> > is way safer than C.
> > 
> > Of course, we would have to keep C for a while, but given that
> > firefox depends on rust and Fedora ships it, gdb support it etc,
> > there is a good chance it ends in every single platform we care
> > about.
> 
> I'm just curios - do you know if it is currently possible to compile
> it using mingw ?

Good point, I haven't looked much at that either. I know the cross-compilation story is fairly good, and you can see answer on SO about cross-windows build http://stackoverflow.com/questions/31492799/cross-compile-a-rust-application-from-linux-to-windows 

> Anyway I believe that is better to have tests before we change
> something :)

sure, I would just prefer we reuse libraries when possible instead of reinventing the wheel, and URI parsing is a good candidate.