Re: Call for port testing on fmgr changes -- Results!

On Fri, 16 Jun 2000, Ryan Kirkpatrick wrote:

> I will download the snapshot today at work (where I have "real"
> bandwidth :) and test things out this weekend. I should have a report by
> Monday. Maybe I will even have a patch that can be safely applied to
> the development source tree. :) TTYL.

Ok, I have tested the new snapshot. I have good news and I have
bad news...
First of all, to build pgsql correctly from the snapshot tarball
(dated Fri, Jun 16th), I had to run a 'make distclean' first. There were
some config.status files laying around that were confusing the configure
run. Don't know if that is just par for pgsql devel snapshots or if there
was mistake somewhere in the building of the snapshot.
Once I got past that, I realized that the only non-fmgr related
Linux/Alpha patch was a adding a single line to the linux_alpha template
file. If I had realized that earlier, I would have submitted that patch
long ago. But it is attached now, probably not even work using 'patch' on
(i.e. easier to hand code it in), and really should not break any other
platforms. :)
Now, for the fmgr rewrite testing... I compiled pgsql w/o any
patches or source code modifications save for the two I mentioned above
(make distclean and added line to linux_alpha template). 'initdb' ran
without problems. Regression tests failed on geometry as expected (off by
one in nth decimal place), but also failed on timestamp, tinterval,
horology, and abstime. These latter failures were characteristic of
running release versions of pgsql w/o the Linux/Alpha patches. Years that
are several centuries off. :(
I tracked the problem down to the following suspect functions. I
found these as they were functions listed in the Linux/Alpha patches (as
needing Datum datatypes for function params), but did not appear to have
been rewritten from the new fmgr (i.e. no PG_FUNCTION_ARGS in
declaration). These functions are all in src/backend/utils/adt/nabstime.c.
abstime2tm
tm2abstime
AbsoluteTimeIsBefore
AbsoluteTimeIsAfter
reltime2tm
If these are converted to the new fmgr format, then I think the regression
tests mentioned above should pass on Linux/Alpha. I think I will leave the
conversion to some one who is more familiar with the new fmgr than me.
Once that is done, and the attached patch is applied to the source
tree, there is a very good chance that pgsql will FINALLY work out of the
box on Linux/Alpha. :)
Let me know when the new fmgr is ready to test again, if you
need any more information on the above, or if there is anything else I can
do help this process. TTYL.

---------------------------------------------------------------------------
| "For to me to live is Christ, and to die is gain." |
| --- Philippians 1:21 (KJV) |
---------------------------------------------------------------------------
| Ryan Kirkpatrick | Boulder, Colorado | http://www.rkirkpat.net/ |
---------------------------------------------------------------------------