#652 closed defect (fixed)
32-bit regression failures
| Reported by: | Mike Taves | Owned by: | strk |
|---|---|---|---|
| Priority: | blocker | Milestone: | 3.4.2 |
| Component: | Default | Version: | main |
| Severity: | Unassigned | Keywords: | 32-bit |
| Cc: |
Description
Revision r3853 introduced 3 regression failures for 32-bit systems, as tested with Debian, Ubuntu and NetBSD. There does not appear to be issues with 64-bit versions of Debian / Ubuntu.
Attachments (2)
Change History (15)
by , 11 years ago
| Attachment: | make_check.txt added |
|---|
follow-up: 4 comment:1 by , 11 years ago
Can you check if the query in http://trac.osgeo.org/geos/ticket/527#comment:3 gives you any record in output ? Absence of any record means that the output from ST_UnaryUnion is correctly noded, and that's what the test for #527 is about.
comment:2 by , 11 years ago
| Milestone: | → 3.5.0 |
|---|
I get 3 regression failures (on winnie Windows 2003 64-bit R2) my 32-bit mingw64w32 gcc 4.5.4. I get a lot more for 64-bit on Winnie (but I think something else is going on with that). On an old windows xp I tested (I think 3.5 doesn't compile at all and that's a recent development with latest commits I think)
I've attached my make check log. Haven't check to see if its the same failures as Greg
by , 11 years ago
| Attachment: | winnie_w32_454makecheck_geos3.5.txt added |
|---|
comment:4 by , 11 years ago
Replying to strk:
Can you check if the query in http://trac.osgeo.org/geos/ticket/527#comment:3 gives you any record in output ? Absence of any record means that the output from ST_UnaryUnion is correctly noded, and that's what the test for #527 is about.
There were 0 rows from the query with 32-bit Ubuntu 13.04, POSTGIS="2.1.0 r11822" GEOS="3.4.1-CAPI-1.8.0 r3892" ...
comment:5 by , 11 years ago
On NetBSD 6.1_STABLE i386, with geos 3.4.1, I get 3 regression failures. On NetBSD 5.2_STABLE amd64, with geos 3.4.1, all regression tests pass.
comment:6 by , 11 years ago
Warning flag here:
../../include/geos/geom/BinaryOp.h:459: warning: overflow in implicit constant conversion
comment:7 by , 11 years ago
| Milestone: | 3.5.0 → 3.4.2 |
|---|---|
| Owner: | changed from to |
| Priority: | major → blocker |
| Status: | new → assigned |
comment:8 by , 11 years ago
| Resolution: | → fixed |
|---|---|
| Status: | assigned → closed |
comment:10 by , 11 years ago
make check of a distfile built from trunk passed on NetBSD 5 i386. (This is like distcheck, but via pkgsrc.)
(I had mumbled about testing on sparc, but the sparc I have is building about 200 .o files per day, and thus it will be several more days before it finished 3.4.1.)
all output from "make check" with r3896