On 21 Feb 2005 15:06, Richard Levitte - VMS Whacker wrote:
> In message <[EMAIL PROTECTED]> on Mon, 21 Feb 2005 15:48:41 +0100 (CET), 
> [EMAIL PROTECTED] said:
> 
> dl> I hope I won't sound too aggressive here...
> dl> 
> dl> Did you even bother to look at 3.7a5? Pretty much every found bug
> dl> and code-shift problem found in a4 has been corrected and a lot of
> dl> stuff has been added (introducing new bugs, of course). All of
> dl> this required some effort, thinking and reorganizing of the code.
> dl> 
> dl> How about making an effort yourself (for example looking at the
> dl> 3.7a5 changelog, verifying the changes made and testing the new
> dl> stuff) instead of pissing on other people's work?
> dl> 
> dl> Again, sorry for the outburst.
> 
> Hmm, OK, I need som clarity here.
> 
> I recall that there were some problems with Gnome and KDE, are those
> no more?  I don't have the time to test with those desktop systems
> myself, so please help me bring clarity in my mind.
> 
> If 3.7a5 is good enough to go to beta or release state, even with the
> stuff I don't check, I'm all for it!
> 
> I'm using CTWM 3.7a5 myself, just as it is without any desktop system,
> and I'm quite happy with it for now.
> 
> There are still issues in the RT database.  I'll take a look at those
> in the days that come, and will see what I can resolve and what can
> wait 'til later.

In November 2003, I repeated my plea from March 2003 regarding a bug
in the then current 3.7 alpha. In summary:

Basically, any key you map in the workspace context becomes unmapped
in every other context. So, lets say you bind the cursor keys to allow
you to switch between workspaces like so:

"Up" =          : workspace : f.upworkspace
"Down" =        : workspace : f.downworkspace
"Right" =       : workspace : f.rightworkspace
"Left" =        : workspace : f.leftworkspace

...then you can't use the cursor keys anywhere except in the workspace
manager. The keys won't even generate KeyPress/KeyRelease events.

Of course, should you chose to map the numeric keys 1-0 to your ten
workspace, then you'd have to patch your software to handle roman
numerals. ;-)

3.6 + my patches work, but none of the 3.7 alphas I've tried do,
including alpha5.

I do not know why, or what patch caused this, but it seems that it's
an interaction between my workspace patch (#1) and later changes. I do
know that Claude changed things that the workspace patch used, and
also the code of the original workspace patch, in his later (#8)
Xinerama patch, so of the opaque #1-#8 changes, it seems reasonable
that it was the Xinerama patch that broke things. (Or, that the
Xinerama patch revealed a latent bug in my code.)

This was discussed in March 2003, and I then brought it up again in
November 2003, again hoping for (Claude's) help resolving it, since I
wasn't sure exactly what he had done in the Xinerama patch, only that
it was fairly extensive changes. I belive I also offered to try to
narrow it down, if I could get patches #2 - #8 as separate patches.

I received no replies.

Since the bug doesn't manifest with my patches applied to 3.6, and it
seems very plausible that it was introduced or triggered in the
Xinerama patch, and the author of that patch so far has not been
willing to help or take ownership of the bug, I see only a few
possible ways to go forward:

1) Someone else takes ownership of the bug, finds it and fixes it.

   This has not happened in the almost two years that have elapsed
   since it was first discovered.

2) We start from 3.6, apply patches until it manifests, and then back
   out the patch that causes the problem until it can be resolved.

   This is what I wanted (and offered) to do, but this requires
   getting access to individual patches, or at least, intermediate
   versions. (Pre and post #8 would be nice, since #8 is the suspected
   culprit/trigger.)

3) Back out the workspace patch.

   Maybe my patch is at fault, and later changes just exposed the
   latent bug.

The problem, as I see it, is that the Xinerama patch (#8) is big,
changed a lot of code, is marked "preliminary" and Claude, who made
it, so far have seemed uninterested in continuing work on it or even
help resolve problems seemingly caused by it.

As for Dan's comments:

Yes, I've looked at alpha5, it's still broken.

I did make an effort myself, both to find out what was wrong myself,
to get the author of the patch my investigations pointed to as the
likely culprit (or at least trigger) to help and to get access to the
individual patches so I could at least trace the bug and make sure
which patch actually triggered it.

All these (repeated) attempts failed.

So, seeing as we have an opaque blob of code containing at least one
fairly serious bug at the start of 3.7 that nobody wants to take
responsibility for, nor seem to be able to untangle the individual
patches for, my suggestion is that we effectively back out that entire
blob.

Since that blob is at the start of 3.7, and subsequent patches are
individually available (and presumbale depend on), it seems easier to
take 3.6 and apply those patches that are know to be good, or for
which there are individuals willing to take stewardship, and then work
on adding whatever patches are left.

Dan, I don't see why you think this is such an unreasonable thing to
suggest, given the circumstances, nor why you feel the need to be
abusive about it.

/Bj�rn

Reply via email to