bug#11883: 24.1.50; shell-mode doesn't colorize the prompt

[Lars Ingebrigtsen <larsi@gnus.org>]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

> The behavior you see is due to the following:
> - the prompt is highlighted with an overlay that uses
>   a comint-highlight-prompt face.
> - "the prompt" is determined dynamically as "the last
>   non-line-terminated output of the process".
> - the ansi-color escapes used to be applied using overlays but are now
>   applied using text-properties.
>
> When you use ansi escapes to color the prompt, you clearly have
> a conflict with the comint-highlight-prompt face that is to be used for
> the prompt.

The following simple tweak seems to fix the reported issue:

diff --git a/lisp/comint.el b/lisp/comint.el
index 9e406614b9..3dfb269de4 100644
--- a/lisp/comint.el
+++ b/lisp/comint.el
@@ -2157,9 +2157,9 @@ comint-output-filter
 		 'comint-highlight-prompt))
 	      (setq comint-last-prompt
 		    (cons (copy-marker prompt-start) (point-marker)))
-	      (font-lock-prepend-text-property prompt-start (point)
-					       'font-lock-face
-					       'comint-highlight-prompt)
+	      (font-lock-append-text-property prompt-start (point)
+					      'font-lock-face
+					      'comint-highlight-prompt)
 	      (add-text-properties prompt-start (point)
 	                           `(rear-nonsticky
 	                             ,comint--prompt-rear-nonsticky)))

That is, we don't overwrite the ANSI face properties.

I guess this might also avoid overwriting properties in prompts that use
ANSI codes?  Which might also be nice.

Would this regress anything?  Anybody got an opinion?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no