HEADSUP: Commits Planned for 64-bit time_t on sparc64

From: Garance A Drosihn <drosih_at_rpi.edu>
Date: Sun, 22 Feb 2004 21:58:45 -0500
   [this is being BCC'ed to -arch and -hackers just to make sure
    that everyone is aware of this before changes are committed,
    but I expect all of the discussion to happen on -current]

Sparc64 users (including me) have said that we'd like the sparc64
port to be running with a 64-bit time_t before we make a "stable"
branch for the 5.x-series.  I have been working on a procedure
that can be used to safely perform this very incompatible change.
I think I finally have that pretty much ready-to-go.

I will therefore make the bold claim that I plan to take the
files you can find at:

http://people.freebsd.org/~gad/time-64/UPDATING.64BTT
http://people.freebsd.org/~gad/time-64/installworld_oldk 
http://people.freebsd.org/~gad/time-64/installworld_newk

and I plan to commit them to /usr/src, on Wed March 3rd.  This
does not actually change anything, but it tells adventurous
sparc64 users that we're officially on the way to making this
change, so they can try upgrades to 64-bTT (64-bit time_t).

These files are only going to be in the base system long enough to
help sparc64 users through the transition.  I am very interested
in any "show-stopper" problems in what I have written.  My hope is
that these files will disappear shortly after 5.3-release, though,
so I don't want to spend much time polishing the scripts up when it
comes to style issues.  I'm still interested in hearing about them,
but I may not be in a rush to do anything about style issues.

Assuming that no one runs into a show-stopper problem, I plan to
commit the change to /usr/src/sys/sparc64/include/_types.h that
switches to 64-bTT for sparc. /usr/src/UPDATING will also need
to have an entry added, but that entry will just point to the
/usr/src/UPDATING.64bTT file.  I plan to commit this (or have
Warner commit the change to UPDATING?) on Wednesday March 10th.

All of these changes will cause zero changes for people who are
running on other hardware platforms.  Only sparc64 users have to lay
awake in terror of what this change might do to them.  I should note
that it *has* been working fine for me...  :-)   Also, about a dozen
other sparc64 users have used these files to make a successful
transition, with very few problems reported.

As for me, I expect that my real-work job is going to get much busier
by March 15th.  If this 64-bTT change is not committed at that point,
then some other developer will have to lead the charge for making
the change.  And, personally, I think we are already awfully close to
the time for 5.3-release, and we can not afford to get much closer to
it before making this change.  So if we can not make this change by
March 15th (at the latest), then I think we will have to put it off
until 6.0.  In fact, if I had a fix for the dhclient issue I would
prefer to move both of the planned commits up by a week.

I know this could be improved upon if I had more time to work on it,
but this is basically the best I could do with the time I had.  If
my main job had something to do with FreeBSD, then I would have some
way to justify spending more time working on issues like this...

-- 
Garance Alistair Drosehn            =   gad_at_gilead.netel.rpi.edu
Senior Systems Programmer           or  gad_at_freebsd.org
Rensselaer Polytechnic Institute    or  drosih_at_rpi.edu
Received on Sun Feb 22 2004 - 17:58:48 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:44 UTC