[m-dev.] release status

Fergus Henderson fjh at cs.mu.OZ.AU
Sat Nov 23 07:54:31 AEDT 2002

Summary: I think we are just about ready to ship it.

There are still some bugs, but I don't know of any regressions
(bugs which were not present in earlier versions of Mercury),
and there are lots of bug fixes in this version.
Eventually you reach the point of diminishing rewards
where it is better to ship it, so that users can take
advantage of the bug fixes that you have made,
and of course of all the new features.

We are now passing (or should tonight pass) all the tests on all of the
Linux machines, including with lcc and with the gcc back-end;
on murlibobo (alphaev5-dec-osf5.1); and on taifun (sparc-sun-solaris2.7).
We also pass almost all the tests pass on Mac OS X.
We do need to test more on Windows, though.

There are a couple of failures on Mac OS X, but these are all due to
unsupported features (interactive queries in mdb, which are not
supported because we don't yet support shared libraries;
and the `--split-c-files' option to mmc).

> The tests on mundroo (i386-pc-solaris2.8) still have a number of failures:
>                 - the interpreter and vqueens tests in
>                   extras/trailed_update
>                   fails in grade asm_fast.gc.tr: "segmentation violation".
>                 - the CLP(R) tests fail in grade asm_fast.gc.tr.debug,
>                   with message "Killed",
>                   after some warnings about "text relocation remains
>                   against symbol"
>                   Maybe due to problem with --pic-reg?

These two are still not fixed.  But Solaris/x86 is not an important platform,
and the extras/trailed_update and extras/clpr packages are not critical.
As far as we know, they are not regressions; that is, this stuff may
have never worked on that platform.  So I don't think they are worth
holding up the release for.

There are also some newly discovered bugs in extras/lex.  It would be
nice if those can get fixed on the release in time, but again these are
not regressions, and packages in extras are not critical.

There are also a number of bugs documented in the BUGS file and/or
in XXX comments in the Mmakefiles in the test suite.  None of those
are regressions, and most of them were present (and known of) when we
shipped 0.10.

>                 - the tests in tests/debugger all fail,
>                   and so do all the tests in debug grades,
>                   due to libreadline.so not being found
>                   (need to link with `-R/usr/local/lib',
>                   or set LD_LIBRARY_PATH?)

That should be fixed now.

There are still a number of bugs which are IMHO not release critical:

 	- the two failures on mundroo mentioned above
 	- some failures
 	- the lex bugs
 	- the bugs mentioned in the BUGS file
 	- the bugs mentioned in XXX comments in
	  the Mmakefiles in the test suite
 	- the bugs mentioned in bug reports filed in

I won't object if someone manages to fix them before the release is
frozen.  But I do not plan to fix these myself.

That leaves the following tasks on my list:

 	- Test on Windows
 	- Test the final source and binary distributions

 	- Update web pages for new release [fjh]
 	- Document MSVC / Mingw installation issue with Cygwin vs Windows
 	  path names
 	- Update docs for .NET back-end (web page; also WORK_IN_PROGRESS?)

So, I propose we do the testing and documentation tasks above,
fix any new release-critical bugs that we discover during this
testing, and then ship it.


Fergus Henderson <fjh at cs.mu.oz.au>  |  "I have always known that the pursuit
The University of Melbourne         |  of excellence is a lethal habit"
WWW: <http://www.cs.mu.oz.au/~fjh>  |     -- the last words of T. S. Garp.
mercury-developers mailing list
Post messages to:       mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions:          mercury-developers-request at cs.mu.oz.au

More information about the developers mailing list