On Sat, Oct 1, 2011 at 13:44, Matthew Brush mbrush@codebrainz.ca wrote:
On 11-10-01 03:05 AM, Enrico Tröger wrote:
On Mon, 26 Sep 2011 01:21:59 -0700, Matthew wrote:
So I was just curious *specifically* what distro(s) and version(s) sets the oldest version of GTK+ that Geany should support?
I don't know anything as specific as you probably want to hear, sadly. It's just that it raises up from time to time when users tell on what old systems they compile Geany.
It seems to me like this group of users would be very small and require that they are: - Enterprise/LTS users running a GUI - And who are willing to install unsupported software. - And whos distro doesn't backport newer libraries. - And who are developers, can compile Geany but can't compile GTK+. - And who are not satisfied with existing supported Geany packages.
I would just be curious to know how many (or what percentage of) users fall into this group.
I belong to this group (at work where we use SLES 9 with GTK 2.4) except that I compile GTK to run Geany. As a rule, I use old versions of programs for about everything except Geany. Because of chain of dependencies, you'd have to compile almost the whole desktop in order to make the program running. With Geany I'm really happy it has GTK dependency only so relatively little has to be compiled - pkgconfig, freetype, fontconfig, glib, pixman, cairo, pango, atk, gtk. Yes, I call it relatively little because things get much worse for other software.
Conclusion: people on enterprise linuxes don't expect to be able to run the latest software because ABIs and APIs change very frequently for various libraries and you can be almost sure that you'll have to recompile lots of system libraries in order to get the particular application running.
Since we recently bumped the GTK requirements to 2.12 for the upcoming release, do we need to discuss this already again? Yeah, newer versions are cooler but is it necessary or just cooler?
It's not so much about "cool" as it is about using improved versions of the libraries. Lots of good stuff has been added since 2.12 (ie. in the last 4+ years of active development).
Just one example would be the GtkBuilder stuff which would not really be "cool", but it would make it easier to unglue some of the business logic from the GUI stuff and make fixing/developing Geany easier. Another example could be the use of GtkInfoBar, whether and where it might be useful is another discussion.
What's more, some existing #if'd code could be fixed to be used proper, again making Geany easier to fix/maintain (assuming there's some existing #if'd code blocks...I think I saw some).
Another reason, is to make it easier for people to help get Geany able to build with GTK+ 3 eventually, which probably many more users will care about in the somewhat near future than supporting extremely old versions of GTK+ 2. It would also help to attract new developers to Geany if they didn't have to "go back in time" to see what was the way to do GTK+ things 4+ years ago. No one likes writing deprecated code :)
IMO, Geany releases should keep up to date with the versions of libraries of mainstream distros (Debian, Ubuntu, Fedora, Etc.). It's not like the older versions of Geany for older GTK+ versions are going to be any less usable over time.
Of course this is just my $0.02 cents as some who has had many good ideas for improvements to Geany that have been stopped with needing to support ancient GTK+/GLib versions.
In general I'm happy Geany is rather conservative about its dependencies however in moments like this, where old development tools stop being supported and more and more linux distributions are moving to GTK 3, I think the dependency bump is the better option. Otherwise it would mean that no work on GTK 3 support could be done in the following year(s). Moreover, if the next version of Geany will be released in say one year, the support of old systems should be about the same as now (with half-year release cycle 2.12 is one year from 2.16).
Cheers,
Jiri