[sword-devel] Does Sword have q/a regression tests?

Chris Little sword-devel@crosswire.org
Wed, 22 Oct 2003 15:45:02 -0500


This isn't a comment on the question at hand itself, since I don't know 
the status of regression tests.  I think we were discussing it at one 
point, but don't know the results of that discussion.

If you think this was a bug (beyond hypotheticality), you should know 
that this is actually ICU code, which we don't really maintain.  (In a 
sense, we maintain our own version, but we mostly just mirror the 
official version of ICU and do a few minor patches to make it run in 
Borland.)  So if you find a bug in this code, it is quite possibly the 
ICU team to whom you should report it.

However, I'm not sure how you're getting this compiler error.  icu-sword 
compiles cleanly for me in both MSVC (6.0 & 7.1) and Borland C++ Builder 
(5.0).  Make sure you're doing all of the conditional defines in your 
project file or makefile. Defining U_HAVE_PLACEMENT_NEW=0 should correct 
this problem.

--Chris


Lynn Allan wrote:

> Inquiring (newbie) minds want to know:
> 
> * Does the Sword project have a suite of tools for regression testing?
> 
> The reason I ask:
> I'm trying to rebuild diatheke and ActiveDiatheke from "scratch", including
> the ICU versions. This involves libsword.icu and libsword6icu.lib. For the
> purposes of discussion, suppose I found what appears to be a bug, using this
> example (just hypothetically, of course)
> 
> uobject.h has multiple definitions of method delete because of
> #define U_HAVE_PLACEMENT_NEW
> which keeps it from compiling.
> 
> Further suppose I made a quick/dirty hack-fix and commented out my best
> guesstimate of the probable extra method. This lets the compile proceed.
> 
> It leads me to wonder: If I submitted this "fix", is there regression test
> suite to run against? If so, is this something a developer has access to?
> I'd like to avoid embarassing myself with a "fix" that breaks other code.
> 
> 
> _______________________________________________
> sword-devel mailing list
> sword-devel@crosswire.org
> http://www.crosswire.org/mailman/listinfo/sword-devel