[3758] in testers
Re: More 8.2 and exmh problems
daemon@ATHENA.MIT.EDU (Jacob Morzinski)
Mon Jul 13 05:02:31 1998
To: bug-sipb@MIT.EDU
Cc: "Roger A. Roach" <rar@MIT.EDU>, testers@MIT.EDU
From: Jacob Morzinski <jmorzins@MIT.EDU>
Date: 13 Jul 1998 05:02:17 -0400
In-Reply-To: "Roger A. Roach"'s message of Sun, 12 Jul 1998 17:25:57 EDT
<daemon@ATHENA.MIT.EDU> (Roger A. Roach) writes:
> I tried to use the alternate editor (emacs) and got:
>
> Error in script: X server insecure (must use xauth-style authorization);
> command ignored
> while executing
> "send $exmh [list Edit_Ident [winfo name .]]"
> (file "/mit/sipb/arch/sun4x_56/bin/exmh-async" line 17)
Short answer: exmh relied on being able to use Tk "send" to
know when the alternate editor was done. This has never worked
SGIs, and apparently won't work on 8.2 Suns.
To still be able to use the alternate editor, your least worst
option is to change the "Preference" section on "Editors", and
tell exmh that the second editor is merely "emacs". (No ampersand.)
This has some drawbacks, mentioned below.
Long answer:
I may need some help on this one. Does anyone have xauth clue?
Exmh tries to use Tk's "send" command to launch background tasks
that report back to the main application when they finish.
Unfortunately, send has never worked on the SGIs, and now it
appears that it won't work on 8.2 Suns either. Therefore, exmh's
interactions with external editors it launches are limited.
My options for dealing with this are:
1) Launch the user's editor as a background process, in which
case control never returns to the exmh application.
This is suboptimal, because it's not obvious how to send your
mail once you finish editing it.
2) Launch the editor, and have exmh wait for the editor to
finish. This has the advantage that control will return to
exmh when the editor finishes, but has the disadvantage that
exmh will freeze up (not respond to events) until that time.
This isn't great, but I could live with it.
3) Find out what happened with xauth authentication. Try to
fix it, at least on Suns. Be nice if it could be made to
work on SGIs too.
I don't have knowledge of how to begin a solution like this.
I'd love to do #3, but think I'll have to do #2.
Any thoughts?
-Jacob Morzinski
