Re: [Announce] WinSANE 0.1.0.0 Release

Nick Lamb (njl98r@ecs.soton.ac.uk)
Mon, 10 May 1999 14:34:52 +0100 (GMT)

On Sun, 9 May 1999, Brian Macy wrote:

> Things I really could use help/info on:
> - Why endian matters when do the actual scan. From everything I can tell the
> scan data is all byte oriented so endianess doesn't matter.

For a bit-oriented scan (lineart) you can safely assume everyone is agreed
about which order the bits go in. Networks transfer a byte at a time.

For a byte-oriented scan (8-bit grey, 24-bit colour) there's also nothing
to worry about, we're all agreed how to send a byte over a network.

However, multi-byte samples (10 bit grey, 36 bit colour etc.) require
that you compare your endianess with the backend, and if they're different
convert. This ONLY applies to image data, everything else uses network
byte order to solve this problem.

If (as I suspect) this version of WinSANE doesn't support anything bigger
than 8-bit samples, (ie 24-bit colour) then you needn't worry about it.

> - TWAIN DLL... I know almost nothing about TWAIN and nothing about
> implementing a TWAIN DLL. I don't have the time to really figure it out
> either so without help it will never progress to be a TWAIN DLL.

Hmm. The TWAIN toolkit, at least in recent versions looked like a pretty
straight forward "fill in the blanks" piece of software. I can't offer to
help here (no Win32 experience, and no desire to get some) but I suspect
that plugging your GUI onto TWAIN is pretty easy.

Since SANE already handles all the fiddly decisions about how to express
scanner capabilities, in the backend, you'll probably have less work to
do than someone implementing TWAIN for a single device... I hope.

> Adds support for LineArt, HalfTone, and GrayScale
> Allows selecting of device on remote host
> Allows selecting of Scan Mode

Hooray!

Nick.

--
Source code, list archive, and docs: http://www.mostang.com/sane/
To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com