Re: [xml] runtest mystery bug: name2.xml error case regression test

[Daniel Richard G. <oss teragram com>]

On Thu, 13 Sep 2012, Daniel Veillard wrote:

> > What I was asking is, should everything still work correctly with
> > that larger value? Not break anything, still give the same results?
> > (Because if not, I'll need to find some other way of making the bug
> > reproducible on FC17.)
>
>  It may change some of the output as you reported, but not the document 
> content, so if that's the way you reproduced it on F17 okay

I'm pretty sure I remember the error messages produced within runtest 
being different from the reference, so if it's possible for 
different-but-still-correct output to be generated, that shouldn't be the 
case here.

> > Well, you know what's going on much better than I do ^_^  Can you
> > reproduce this?
>
>  No, I don't have a precise idea of what you actually changed nor which 
> file exposes the problem nor how you reproduce it except by running the 
> modified runtest.c , does that show up in xmllint ?

I wouldn't know what to do with xmllint :]  I just hack on this code, I 
don't actually use it! (I.e. I'm doing release engineering here, not 
application development)

> > The printf()s I alluded to earlier went into that conditional,
> > showing which branch was followed: "base has changed" or "base is
> > equal". With the cut-down runtest, when it succeeds, there are eight
> > "equal" lines and one "changed". When it fails, there are nine
> > "equal"s. So it seems like a *lack* of growing is what leads to the
> > bug....
>
>  And the big buffer might be one of the way to force that behaviour.
> Can your git diff your setup and tell me how you reproduce ?

This is all it is:

--- a/include/libxml/parserInternals.h
+++ b/include/libxml/parserInternals.h
@@ -86,7 +86,7 @@ XMLPUBVAR unsigned int xmlParserMaxDepth;
  * The parser tries to always have that amount of input ready.
  * One of the point is providing context when reporting errors.
  */
-#define INPUT_CHUNK    250
+#define INPUT_CHUNK    250000

 /************************************************************************
  *                                                                     *

The runtest.c patch is optional. With the above, runtest breaks on FC17.


--Daniel


--
Daniel Richard G. || danielg teragram com || Software Developer
Teragram Linguistic Technologies (a division of SAS)
http://www.teragram.com/