Re: Writing Fujitsu M3091 backend

From: mh (crapsite@gmx.net)
Date: Tue Feb 27 2001 - 11:04:30 PST

  • Next message: Frederik Ramm: "Duplex (was: Fujitsu M3091 backend)"

    Frederik Ramm, Die, 27 Feb 2001 :
    > Hi,
    >
    > > When scanning the first page, half the data goes in this large buffer,
    > > then when the frontend asks for another page it comes out of the
    > > buffer very quickly. The buffer can be discarded when the frontend
    > > calls ..._close() so it won't take up RAM all the time.
    >
    > That sounds reasonable. But can I be sure that front-ends will support
    > reading multiple pages one after the other with no call to ..._close()
    > in between? This would not, for example, work with "scanimage", would
    > it?
    You can't be sure, but you can be quite sure that at least the graphical
    frontends won't call sane_close() between scans.
    On the whole, it's just an insufficiency in the current SANE Standard.
    Probably a simple
    SANE_Bool another_scan_follows
    in SANE_Parameters would be sufficient. If another_scan_follows is true
    and last_frame is true, the frontend knows, that the next frame belongs to a
    new image. This, of course, will only fly if the backend stores the
    following frames in RAM, like suggested by Nick Lamb.
    Another advantage is, that this doesn't break compatibilty with
    current frontends. They just read one image and call sane_cancel(); now
    the backend knows, that it can discard the following frames.

    Just MHO
    Michael

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



    This archive was generated by hypermail 2b29 : Tue Feb 27 2001 - 11:34:11 PST