Re: XML_SetHashSalt(p, 0) semantics

How big of an issue is this?
I'd rather fix the docs (which can be viewed online without a new release) than make another release for which I won't have time for a while.

One could also say that it is a good thing that test cases are forced to be fixed....

Once we will do another release, would there still be a need for changing this?
Maybe use another sentinel value, like 0xFFFFFFFF?

Karl

-----Original Message-----
From: Joe Orton [mailto:[email protected]] 
Sent: May-14-12 12:10 PM
To: Karl Waclawek
Cc: [email protected]; 'Kurt Seifried'; 'Tomas Hoger'; 'David Malcolm'
Subject: Re: [Expat-discuss] XML_SetHashSalt(p, 0) semantics

On Fri, May 11, 2012 at 01:16:15PM -0400, Karl Waclawek wrote:
> Yes, this should have been documented. Please file a bug report.

https://sourceforge.net/tracker/?func=detail&aid=3526620&group_id=10127&atid=110127

> 0 is used as a sentinel value that means that no secret salt value has 
> been set and therefore one needs to be generated.
> 
> In any case, one  can always use another value if a fixed hash salt is 
> desired, but I wonder why one would want to turn this off.

My guess: it's useful because the hash ordering randomisation fixes have a tendency to break test cases which make bad API assumptions; allowing a zero hash (and hence the old hash ordering) is at least a temporary reprieve until those test suites are fixed.

Regards, Joe