Remember the tired old argument against using open-source software for meaningful projects? It goes something like this:
You can’t rely on open-source developers because they’re only in it to ’scratch their itch’ and there is no attention to quality beyond their own personal interest
That line used to ruffle my feathers. Seemed like a throw-away line from commercial players who didn’t like the extra competition. I know that in my projects, I do everything I can do maintain backwards compatibility, fix all known bugs, and QA everything before I release. And I’m a one-man show. Well now, the Gtk+ team are doing what they can to prove these commercial players right, blighting other open-source projects, and adding fuel to the commercial nay-sayers’ nay-saying.
In a classic move, best known as ’shooting oneself in the foot’, they’ve broken their own GUI builder, Glade. Here’s the proud release announcement:
I am proud to present to you Glade 3.6.7: the “Horizontally Oriented”
Release, which is yet another bugfix release on the ever so stabler
3.6 series.This release will make everything horizontally oriented in any project
created prior to GTK+ 2.16. Thats right, most projects when loaded
in Glade will be horizonally oriented (vbox orientations will be horizontal).To fix this in your pre 2.16 created project, you must set the
orientation of your vboxes manually and save.
Hey. Smooth move. So all my existing projects ’simply’ must be opened, and all vboxes must be edited. Lucky I don’t use vboxes FUCKING EVERYWHERE!
I’m not the only one to report such a bug, which is prompty marked as a duplicate of http://bugzilla.gnome.org/show_bug.cgi?id=587256. As I ( politely – far more so than here ) pointed out in that bug, if this were my workplace, and we’d created a bug which was now going to fuck over all our customers, we’d do what it took to correct things. What’s more we’d do it before releasing, not after. We certainly wouldn’t tell customers to open all their projects and piss around with the settings of every second object. Finally, we wouldn’t make snide comments that suggest that if customers are so upset about the fuckup, then maybe they should fix it themselves! But that’s precisely advise and sentiment the gtk user community are receiving from the developers.
Call me old-fashioned (and hey, I just made it to 33, maybe I’m getting long in the tooth ), but this isn’t how you treat people if you want them to stick around.
But wait! The poor, poor Gtk+ developers are too few, too hard worked, and too unloved for this to be fair criticism, right? Christ no. Just like the commercial nay-sayers say, they’re just all too introverted and disinterested to attend to anything other than their own itching balls. If there are too few of them, it’s because they keep alientating developers.
Take me for example. No, I’m not talking about this particular fuckup. Ages ago, I reported a critical dataloss bug with GtkComboBoxEntry widgets. Luckily, there don’t appear to be any mission-critical applications using Gtk+, by the way. Not a lot happened. Fine. I’d already done a work-around in Gtk2::Ex::DBI, but it was clearly in the wrong place – sitting right up the top in Perl, where it really should have been in Gtk+ itself. For large models, it would be quite slow. What’s more, I could only really fix GtkComboBoxEntry widgets that were used directly by Gtk2::Ex::DBI itself, which seemed less than satisfactory. So I bit the bullet and spent quite a few hours hacking on Gtk+. This was my first attempt at C. I finally got it to a point where it looked like it was working, and submitted a patch. The next day I managed to fix some warnings, and documentation, and generally clean things up. A Gtk+ developer responded, saying that it appeared to work, but had incorrect indentation, and would probably slow things down with large models ( unavoidable ). But despite the fact that this dataloss bug was now fixed, nothing happened AT ALL for over a year. Oh yeah – I also posted to the Gtk+ mailing list, asking for someone to review it.
So fast-forward to now. I report the our latest bug, and get told that if it upsets me so, then maybe I should fix it myself. Sure, that’s what I feel motivated to do. So I can sit and watch this patch rot for years too. No, fuck that. If there is another so-called “stable” release of Gtk+ with this bug remaining, I will be liberating my code of Gtk+, and Qt is looking better each day.