Opened 19 years ago
Closed 14 years ago
#906 closed defect (fixed)
Byte-order sensitivity leads to bus errors on Macs when reading astrometry catalogs
| Reported by: | Owned by: | heather | |
|---|---|---|---|
| Priority: | normal | Milestone: | |
| Component: | DVO | Version: | unspecified |
| Severity: | normal | Keywords: | |
| Cc: | martin@… |
Description (last modified by )
Created an attachment (id=104)
Paul's patch to address the byte-order issue
Paul has already created a fix addressing the bug (see attachment), but since Nicolas asked me to make sure it's in the next IPP release, here's a PR to trace the issue.
The way Paul described it, this issue should just occur on PowerPCs, but Nicolas sounded as if it also occurred on Intel. Maybe there's a second byte-order issue somewhere?
Not sure what the psLib version or "component" is.
Attachments (1)
Change History (8)
by , 19 years ago
| Attachment: | psMetadataConfig_endian.patch added |
|---|
comment:1 by , 19 years ago
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
Bug has been fixed in CVS head; it will definitely be in the next release.
I'd be interested in hearing of any endian problems with little-endian machines; we use these for our development, and I would have thought we have them under control.
comment:2 by , 19 years ago
| Resolution: | fixed |
|---|---|
| Status: | closed → reopened |
there was some mis-communication on this: paul's patch referred to an endian problem which he and Micheal tracked down in psLib. Nicolas' problem, however, comes from Ohana failing to catch the correct endian-ness of one of the two darwin platforms. I have not fixed this in the Ohana tree yet; Paul's statement (fixed in head) referred to the other problem in psLib.
comment:3 by , 19 years ago
To be a bit more precise, on an Intel darwin (MacOS X.4.10, x86, ipp-2.2) I get:
[panapple:PanStarrs/tests/simtest] martin% getstar -v -D CATDIR /Volumes/Data/ipp/catdir.synth.bright
-region 0.0 0.0 1.0 1.0 -o test.fits
opening test.fits
creating new file test.fits
opening /Volumes/Data/ipp/catdir.synth.bright/fullsky.cpt
no stars in catalog, skipping
SUCCESS
The catalogue doesn't seem to be read and I tracked that to the binary file Photcodes.dat in the catalogue (not sure it comes from there though).
comment:4 by , 19 years ago
| Owner: | changed from to |
|---|---|
| Status: | reopened → new |
comment:5 by , 17 years ago
| Component: | astro → Ohana |
|---|---|
| Owner: | changed from to |
| Severity: | critical → major |
| Status: | new → assigned |
the ohana byte-swap ifdef was missing darwin_x86. it now has it, so this should be resolved.
comment:6 by , 16 years ago
| Component: | Ohana → DVO |
|---|---|
| Description: | modified (diff) |
| Owner: | changed from to |
| Priority: | high → normal |
| Severity: | major → normal |
i'm assigning this bug to heather : can you confirm that a DVO database written on linux can be read on Mac OS/X darwin_x86?
comment:7 by , 14 years ago
| Resolution: | → fixed |
|---|---|
| Status: | assigned → closed |
I can read dvodbs created in linux on my mac (I just tried with one of the CNP.V2.MINIs). It worked fine.
Paul's patch to address the byte-order issue