Re: Undeprecate GtkAlignment ?

[Niels Nesse <nnesse sonic net>]

On 06/04/2015 05:02 AM, Emmanuele Bassi wrote:
> Hi;
>
> On 4 June 2015 at 12:42, Niels Nesse <nnesse sonic net> wrote:
>
> > > Deprecated classes, properties, signals, and functions in GTK+ do not
> > > go away until the major version of GTK+ is bumped. At that point, you
> > > can keep targeting the previous major version of GTK or port to the
> > > new major version, alongside with every other API changes we made in
> > > between. It's important to note that we're not planning a major
> > > version bump any time soon.
> >
> > I wasn't confused about the deprecation model. My question was more
> > about what the consequence of the eventual removal might be. How would
> > an application go about getting the same effect at GtkAlignment without
> > using it directly? Again assuming values for scale and alignment that
> > are fractional.
>
> The proper way would be to use their own custom widget for that.
>
> If the licensing terms allow it, you could even copy-paste
> GtkAlignment in your project.
>
> Alternatively, you bundle your own version of GTK+ and never update
> it; or you never update to GTK+ 4.0 until you decide to redesign your
> UI.
>
> As I said, I personally think that using fractional alignments is
> wrong and just a way to get around the fact that using pixel-precise
> positioning is not something GTK recommends. When we implemented the
> alignment properties for GtkWidget before releasing 3.0 we looked at
> what people were using in open source projects, and we found that the
> vast majority were using 0, 0.5, and 1.0 as alignment values — and the
> few people that required different alignments were better served by
> padding, margin, and spacing dependent on font size, not on fractions
> of the parent widget's size.
>
> Ciao,
>  Emmanuele.

That clears things up. I don't have any code that depends on a
particular use of GtkAlignment BTW. I was just was wary about losing a
possible tool. Since I can get similar functionality in a pinch in
custom code I'll avoid addressing the issue in bug #745317.

Thanks,
Niels