Created attachment 129785 [details] [review]
patch

i can't get gnome-mag to *fail* when compiling, even by setting complete bogus module import names and using -Wl --as-needed. Frustrating.
Hence this patch is *completely* untested and probably wrong.

We should probably get rid of gnome-mag - the code looks... ancient.

...or gtk.show_uri ? shrug

It wouldn't be gnome-mag but the colorblind applet, but it won't add to my panel, even after I fixed an obvious bug:

ColorblindApplet.py has:
  from colorblind.ui.ColorblindPreferencesUI import show_preferences

but ui/ColorblindPreferencesUI.py has:
  import colorblind, colorblind.Utils, colorblind.ColorblindApplet

which means the import of that file will stop before show_preferences is defined, which in turn means the import in ColorblindAppley.py fails with ImportError.

Created attachment 131118 [details] [review]
Replaces gnomevfs.url_show by gtk.show_uri

Patches are good to commit

Thanks

Reassigning bug to Willi, as Bill no longer work for Sun. Willi, request to assign the bug to concern developer.

Thanks

(In reply to comment #6)
> Reassigning bug to Willi, as Bill no longer work for Sun. Willi, request to
> assign the bug to concern developer.
> 
> Thanks

This definitely needs some love. I spent all day yesterday and all morning today trying to figure out how to get gnome-mag to build.  I'm missing something, and I'm hoping maybe Carlos can help.

(In reply to comment #7)
> I spent all day yesterday and all morning today trying to figure out 
> how to get gnome-mag to build.

What issues are you exactly running into?

(In reply to comment #8)
> (In reply to comment #7)
> > I spent all day yesterday and all morning today trying to figure out 
> > how to get gnome-mag to build.
> 
> What issues are you exactly running into?

Heh - I was just going to explain that.  I can get it to build just fine without libcolorblind, but as soon as I start enabling libcolorblind (which is what I believe this bug is related to), I run into a whole bunch of Python issues.  I believe it may be because I'm working with Python 2.4 and it might be that the gnome-mag stuff requires Python 2.6 (but I'm not sure).

So it seems we talk about the flag '--enable-colorblind-applet=yes' here.

(In reply to comment #10)
> So it seems we talk about the flag '--enable-colorblind-applet=yes' here.

Yep - it's automatically enabled if you have built/installed libcolorblind from http://alioth.debian.org/frs/download.php/1952/colorblind-0.0.1.tar.gz.

Since the patch in question deals in the colorblind subdirectory of gnome-mag, I assumed I needed libcolorblind installed in order to be able to build/test the patch attached to this bug.  The unfortunate thing is that my machine doesn't seem to be up to the Python version needed to make it work.  :-(  I also cannot upgrade my Python to 2.6 right now.  :-(  So, I'm kind of useless when it comes to this patch.

Python 2.5.2 works fine here, but as I said: I can also remove the gnomevfs import and jhbuild still does not fail. That sucks for testing.

(In reply to comment #12)
> Python 2.5.2 works fine here, but as I said: I can also remove the gnomevfs
> import and jhbuild still does not fail. That sucks for testing.

Yeah.  I'm stuck at Python 2.4 right now.  :-(  Let's see if Carlos has some cycles to take a quick look at this.  Carlos, if you can review the patch and approve it, I'll be happy to do the rest (i.e., commit it, spin up a new release, etc.)

For me the patch looks OK and due the comments I guess that it's also works.

I don't know how to solve the concerns presented by Klapper about gnome-mag and jhbuild. If someone can point me to a python code that fails due missing of imports I will be happy to make a patch for gnome-mag fail while running under jhbuild.

(In reply to comment #14)
> I don't know how to solve the concerns presented by Klapper about gnome-mag and
> jhbuild. If someone can point me to a python code that fails due missing of
> imports I will be happy to make a patch for gnome-mag fail while running under
> jhbuild.

For me it doesn't fail. In fact I'd LOVE to get it to fail for testing. :-P
It fails for Willie.

Soooo.... Anybody willing to commit this (and maybe branch for gnome-2-26 before, to be on the safe side) so it can get some more testing?

(In reply to comment #15)
> For me it doesn't fail. In fact I'd LOVE to get it to fail for testing. :-P
> It fails for Willie.
> 
> Soooo.... Anybody willing to commit this (and maybe branch for gnome-2-26
> before, to be on the safe side) so it can get some more testing?

Well...it just fails for me because I'm at Python 2.4.  I have a spare machine laying around that I can try to configure for this, but if someone beats me to it first, I won't complain.

I built and tested this on GNOME 2.26.0 with colorblind-0.0.1 installed (I needed to open/fix bug #577937 in order to test the changes).  To test these changes, I ran the colorblind applet and made sure I did the appropriate actions in the applet's GUI to exercise the code paths (e.g., looking at About, clicking on e-mail addresses, clicking on web page URL's).  These all worked fine, so I committed the patch.

I'll try to spin up a new release for 2.26.1 sometime next week -- the machine I'm used for this ends up creating a configure file that refers to a version of the l10n tools that are not compatible with GNOME 2.24 systems.  The machine I will use to make the release will produce a configure file compatible with both 2.24 and 2.26 systems (I hope).  This is just anal-retentiveness just in case someone wants to build the release on 2.24.

I'm going to keep this bug open as a reminder that a new release with this fix still needs to be created.

Released in gnome-mag 0.15.6, which is good for GNOME 2.24.x and GNOME 2.26.x.