11:31 a.m.
Hi all
There is a bug in htmlize which prevents #'htmlize-region-for-paste from
working properly. Markers are added to a plist using, e.g.
(plist-put places 'head-end (point-marker))
Our documentation clearly states
,----
| use `(setq x (plist-put x property value))' to be sure to use the new value
`----
However, it also says:
,----
| PLIST is modified by side effect.
`----
which doesn't seem to happen. Patch attached
Index: htmlize.el
===================================================================
RCS file: /cvsroot/xemacs/XEmacs/packages/xemacs-packages/text-modes/htmlize.el,v
retrieving revision 1.15
diff -u -r1.15 htmlize.el
--- htmlize.el 6 Jan 2009 08:30:26 -0000 1.15
+++ htmlize.el 15 Oct 2010 10:25:42 -0000
＠＠ -1453,7 +1453,7 ＠＠
(format "<!-- Created by htmlize-%s in %s mode. -->\n"
htmlize-version htmlize-output-type)
"<html>\n ")
- (plist-put places 'head-start (point-marker))
+ (setq places (plist-put places 'head-start (point-marker)))
(insert "<head>\n"
" <title>" (htmlize-protect-string title)
"</title>\n"
(if htmlize-html-charset
＠＠ -1464,12 +1464,12 ＠＠
htmlize-head-tags)
(htmlize-method insert-head buffer-faces face-map)
(insert " </head>")
- (plist-put places 'head-end (point-marker))
+ (setq places (plist-put places 'head-end (point-marker)))
(insert "\n ")
- (plist-put places 'body-start (point-marker))
+ (setq places (plist-put places 'body-start (point-marker)))
(insert (htmlize-method body-tag face-map)
"\n ")
- (plist-put places 'content-start (point-marker))
+ (setq places (plist-put places 'content-start (point-marker)))
(insert "<pre>\n"))
(let ((insert-text-method
;; Get the inserter method, so we can funcall it inside
＠＠ -1519,9 +1519,9 ＠＠
;; Insert the epilog and post-process the buffer.
(with-current-buffer htmlbuf
(insert "</pre>")
- (plist-put places 'content-end (point-marker))
+ (setq places (plist-put places 'content-end (point-marker)))
(insert "\n </body>")
- (plist-put places 'body-end (point-marker))
+ (setq places (plist-put places 'body-end (point-marker)))
(insert "\n</html>\n")
(when htmlize-generate-hyperlinks
(htmlize-make-hyperlinks))
--
note that "property" can also be used as syntactic sugar to reference
a property, breaking the clean design of verilog; [...]
(seen on http://www.veripool.com/verilog-mode_news.html)
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
8:30 a.m.
Marcus Harnisch writes:
However, it also says:
,----
| PLIST is modified by side effect.
`----
which doesn't seem to happen.
You will be able to see the side-effect only if the key already exists
in the plist. In that case you get
places
|
V
('some-key 'some-value 'head-start #<marker at ...> ...)
But if it doesn't already exist, then you get
places
|
V
('head-start #<marker at ...> 'some-key 'some-value ...)
In theory we could put 'head-start at the end efficiently, but I guess
it's done this way on the theory that it's better to have the new
element at the start of the list as it's likely to be accessed soon.
I'll think about changing this in future versions of XEmacs (obviously
for backward compatibility a patch to htmlize is still necessary).
Patch attached
Thanks! I'll try to get to this shortly.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
1:13 p.m.
"Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
In theory we could put 'head-start at the end efficiently, but I
guess
it's done this way on the theory that it's better to have the new
element at the start of the list as it's likely to be accessed soon.
I'll think about changing this in future versions of XEmacs (obviously
for backward compatibility a patch to htmlize is still necessary).
As far as I am concerned, don't bother unless perhaps FSF people did
things differently and code starts relying on this.
The current behavior of the plist functions is documented, yet I am
curious how this could have possibly worked in the past.
It should perhaps be made clearer in this and similar docstrings that
the side effect modification really means that PLIST might have
changed in unforseeable ways after the operation and that it does not
replace proper assignment.
Thanks
--
Marcus
note that "property" can also be used as syntactic sugar to reference
a property, breaking the clean design of verilog; [...]
(seen on http://www.veripool.com/verilog-mode_news.html)
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
2:10 p.m.
Marcus Harnisch writes:
It should perhaps be made clearer in this and similar docstrings
that
the side effect modification really means that PLIST might have
changed in unforseeable ways after the operation and that it does not
replace proper assignment.
In Lisp, "change by side effect" always means "this is an optimization
that reuses existing conses, and potentially has unexpected
consequences if you have variables that contain subobjects of this
object." Variables containing a list structure must always be
refreshed by set'ting them after such operations.
There are a couple score such functions in XEmacs proper, and probably
as many more in cl.el and friends. I don't think such basic
information should be repeated so many times in the docstrings,
although perhaps there should be a manual node on it. (I'm happy to
listen to arguments otherwise, but that's how I think about it now.)
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
4933
days inactive
4936
days old
3 comments
2 participants
participants (2)
-
Marcus Harnisch -
Stephen J. Turnbull