Two minor bugfixes

Martin Stein <[email protected]> Wed, 11 Feb 2015 23:47:09 +0100

Newsgroups	gmane.emacs.cedet
Message-ID	<[email protected]>

First bug:
File: 0001-Check-that-bounds-is-non-nil-in-semantic-ia-complete.patch

In semantic-ia-complete-symbol-menu in ia.el a check whether bounds are 
valid is missing. There is a check in semantic-ia-complete-symbol. The 
bug is triggered when completing something like
  myfun(
with empty prefix (see example in commit message).
Remark: In contrast to the code snippet in semantic-ia-complete-symbol I 
have used a let instead of having four (oref a bounds) evaluations. I 
also omitted the (goto-char (car (oref a bounds)), as delete-region 
already places point at this position.



Second bug:
File: 0001-Save-global-variable-wisent-lookahead-before-calling.patch

The wisent parser uses some global variables, which is a bit of a hassle 
considering that it calls itself (via EXPANDFULL and the like). If the 
last token of a recursive call does not match, it can corrupt the 
unmatched-syntax-cache or even lead to wrong parse results. If recursive 
calls of the parser are only done with parenthesised blocks and rules 
are accordingly, then the last token must be something like ')' and thus 
always matches and the bug is not triggered. For languages like fortran 
this is different.
I am very sure that the commit indeed fixes a bug and nothing more, but 
the wisent parser and its token stream organisation is a rather complex 
automaton, so I might have missed something.

------------------------------------------------------------------------------
Dive into the World of Parallel Programming. The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net/

_______________________________________________
Cedet-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/cedet-devel

0001-Check-that-bounds-is-non-nil-in-semantic-ia-complete.patch (text/x-patch, 1.3 KB)

>From be8a53f750238b0cba5177671f482ceec78656fd Mon Sep 17 00:00:00 2001
From: Martin Stein <[email protected]>
Date: Wed, 11 Feb 2015 23:27:03 +0100
Subject: [PATCH] Check that bounds is non nil in
 semantic-ia-complete-symbol-menu

Delete region fails if bounds are nil.
Bunds are nil if we are completing an empty prefix string,
e.g. type constrained within a function argument list.
Example (in C)

void myfun(int x)
{
...
}

int main()
{
  int locvar;
  myfun(*try to complete here choosing locvar*
}
---
 lisp/cedet/semantic/ia.el | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/lisp/cedet/semantic/ia.el b/lisp/cedet/semantic/ia.el
index 67fc809..e6d71ef 100644
--- a/lisp/cedet/semantic/ia.el
+++ b/lisp/cedet/semantic/ia.el
@@ -188,7 +188,11 @@ Completion options are calculated with `semantic-analyze-possible-completions'."
 	(when ans
 	  (if (not (semantic-tag-p ans))
 	      (setq ans (aref (cdr ans) 0)))
-	  (delete-region (car (oref a bounds)) (cdr (oref a bounds)))
+          (let ((bnds (oref a bounds)))
+            ;; bounds could be nil if we are completing an empty prefix string
+            ;; (e.g. type constrained within a function argument list)
+            (when (and (car bnds) (cdr bnds))
+              (delete-region (car bnds) (cdr bnds))))
 	  (semantic-ia-insert-tag ans))
 	))))
 
-- 
1.9.1

0001-Save-global-variable-wisent-lookahead-before-calling.patch (text/x-patch, 2 KB)

>From cd7567a61566e375a9186345b03938469cfb22e8 Mon Sep 17 00:00:00 2001
From: Martin Stein <[email protected]>
Date: Mon, 19 Jan 2015 22:26:02 +0100
Subject: [PATCH] Save global variable wisent-lookahead before calling semantic
 action

The semantic action might be an EXPAND, EXPANDFULL or some similar
action calling the parser recursively. If the last token
of the recursively called parsing step failed to match it ends up
in wisent-lookahead, and function wisent-parse-stream pushs
it back onto wisent-lex-istream. From there on it simply ends up
occuring twice on the unmatched-syntax-cache, or worse might even
lead to parse errors in rare cases.
---
 lisp/cedet/semantic/wisent/wisent.el | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/lisp/cedet/semantic/wisent/wisent.el b/lisp/cedet/semantic/wisent/wisent.el
index 3278838..fb98c7a 100644
--- a/lisp/cedet/semantic/wisent/wisent.el
+++ b/lisp/cedet/semantic/wisent/wisent.el
@@ -376,7 +376,7 @@ automaton has only one entry point."
          (wisent-parse-lexer-function lexer)
          (wisent-recovering nil)
          (wisent-input (wisent-parse-start start starts))
-         state tokid choices choice)
+         state tokid choices choice wisent-lookahead-save)
     (setq wisent-nerrs     0 ;; Reset parse error counter
           wisent-lookahead nil) ;; and lookahead token
     (aset stack 0 0) ;; Initial state
@@ -468,7 +468,12 @@ automaton has only one entry point."
        ;; Reduce by rule (call semantic action)
        ;; -------------------------------------
        (t
+        ;; save the global variable wisent-lookahead,
+        ;; as the semantic action might be an expand or expandfull
+        ;; which calls this function recursively
+        (setq wisent-lookahead-save wisent-lookahead)
         (setq sp (funcall wisent-loop stack sp gotos))
+        (setq wisent-lookahead wisent-lookahead-save)
         (or wisent-input (setq wisent-input (wisent-lexer))))))
     (run-hooks 'wisent-post-parse-hook)
     (car (aref stack 1))))
-- 
1.9.1