home *** CD-ROM | disk | FTP | other *** search
- To: tiff@sgi.sgi.com
- Subject: Libtiff, DOUBLE and FLOAT
- Date: Mon, 14 Nov 1994 17:56:35 +0100
- From: ndr@tazboy.JPL.NASA.GOV (Niles Ritter)
-
- I've been working with some other people on developing new
- TIFF tags (for storing Geographic Projection info) and we have
- found need for the TIFF 6.0 DOUBLE format tags (SRATIONAL isnt enough).
- However, on looking through the libtiff code I noticed several
- things:
-
- 1) The DOUBLE format is not fully supported: In the routine
- TIFFFetchData() where bytes are supposed to be swapped on machines with
- byte-order opposite that of the file, the FLOAT type is correctly
- swapped, but the DOUBLE is not (I think appending a TIFF_DOUBLE case
- after TIFF_RATIONAL will do the trick, since the swap-procedure is the same).
- This has never been noticed because none of Sam's Tags are IEEE.
-
- 2) Also, when writing DOUBLE's there is no TIFFWriteDoubleArray()
- routine to handle the tag. This routine should in turn make a
- reference to a TIFFCvtNativeToIEEEDouble(), which brings me to
- the next item:
-
- There is no DOUBLE support in the library. Anything that is there
- is unintentional and likely incomplete.
-
- 3) The FLOAT tag is no longer handled for VAX (or a "tahoe", whatever
- that is). Formerly the module "tif_machdep.c" had an IEEEFLOAT2NATIVE
- macro for converting from IEEE to the perverse VAX formats.
- Theoretically, these should have been migrated over to the tif_vms.c
- module, along with the routine TIFFCvtNativeToIEEEFloat() (which gets
- #defined to no-op for IEEE machines, but it looks like they never
- made it. The same holds for DOUBLE.
-
- tif_machdep.c snuck into a couple of distributions by mistake; it's
- been gone from my source tree for quite a while. The code that was
- there was not well tested and incomplete. At some point I decided
- that machines that didn't support IEEE floating point were not interesting.
-
- Has anyone else implemented these tag types, yet? On the VAX?
- If not, I'll probably wind up doing it, and will send the diffs
- to Sam.
-
- Feel free to send me diffs. There should be a new distribution "soon"
- (within weeks).
-
- Sam
-
-
-