| From: | "Jeremy Ford" <jeremford(at)gmail(dot)com> |
|---|---|
| To: | pgsql-bugs(at)postgresql(dot)org |
| Subject: | BUG #4880: gcc (4.4) + double precision differences |
| Date: | 2009-06-25 13:32:27 |
| Message-ID: | 200906251332.n5PDWRV1035801@wwwmaster.postgresql.org |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-bugs |
The following bug has been logged online:
Bug reference: 4880
Logged by: Jeremy Ford
Email address: jeremford(at)gmail(dot)com
PostgreSQL version: 8.3.7 + 8.4.RC2
Operating system: Linux
Description: gcc (4.4) + double precision differences
Details:
I've just had a problem restoring a table on a 8.4.RC2 database compiled
with GCC 4.4.0. The original table (from a 8.3.7 GCC 4.1.2 database) had
some "double precision" values that should have been outside the "allowed
ranges" (values created with PL/R). The pg_restore process failed to
restore any data to that table as (at least) one row had an invalid value:
EST ERROR: "5.926613861253257e-319" is out of range for type double
precision
I have run the following on a few various databases I have access to:
select 1e-319::double precision;
8.2.5 - 32 bit linux (Fedora 8) GCC 4.1.2 --> "ERROR: type "double
precision" value out of range: underflow"
8.3.7 - 32 bit linux (Fedora 8) GCC 4.1.2 --> 9.99988867182683e-320
8.3.7 - 64bit linux (Fedora 8) GCC 4.1.2 --> 9.99988867182683e-320
8.3.7 - 64bit linux (Fedora 11) GCC 4.4.0 --> ".. is out of range for type
double precision"
8.4.rc2 - 32 bit linux (Fedora 10) GCC 4.3.2 --> 9.99988867182683e-320
8.4.rc1 - 32 bit linux (Fedora 11) GCC 4.4.0 --> ".. is out of range for
type double precision"
Besides fixing the data, is there anyway of recompiling with GCC 4.4.0 to
allow these values? Any other options or recommendations?
Cheers, Jeremy.
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Simon Riggs | 2009-06-25 13:33:29 | Re: BUG #4879: bgwriter fails to fsync the file in recovery mode |
| Previous Message | Greg Stark | 2009-06-25 13:01:42 | Re: [BUGS] Integrity check |