backends correctly even if only included files (e.g. headers) are
changed. Don't be too verbose when running makedepend.
Create links libsane-*.so.1 to the real library files if the
links are not there. This fixes dynamic loading on OpenBSD. The
links are not created for MacOS X as they don't work there.
Remove any libsane.* links in /usr/local/lib/sane. Such links
are created by libtool. As they point to libsane-v4l.so,
scanimage -L doesn't find any scanner in case of ld.so
misconfigurations.
The install target is much quiter now and prints the libtool
message only once now.
patches have been applied yet, but it's a start. The backends
are not installed as shared libs but as add-ons, which are shared libs
but in a specific folder in [/system|~]/config/add-ons/ named SANE/.
The install target has been changed to account for this,
Some files don't exist in BeOS/ZETA (ipc.h, ...).
Bbackend function names get a prefix, I suppose to avoid namespace clashes.
Though we do have a libdl to implement dl_open, the native way is prefered.
Added a --with-docdir= configure arg.
BeOS has a broken get[name|addr]info() from bind. This is fixed in ZETA R1.
libtool needs -no-undefined. Various VPATH fixes.
Check for <be/drivers/USB_scanner.h>, but not used yet.
No S_IFSOCK (sockets are fds to /dev/net/api). Stub pio code, untested.
Patch from Francois Revol <revol@free.fr>.
-Detects endpoints for different configurations, interface and
altsetting numbers > 0
- Routine added for changing configuration, interface or altsetting
number
sanei_usb_set_configuration
sanei_usb_claim_interface
sanei_usb_release_interface
sanei_usb_set_altinterface
* backend/mustek_pp.c, backend/mustek_pp.conf, sanei/sanei_pa4s2.c,
include/sane/sanei_pa4s2.h, doc/sane-mustek_pp.man: added global
option "no_epp" to work around a known bug in the linux parport
code.
* sanei/sanei_usb.c configure.in: use usb_interrupt_read instead of
usb_bulk_read for sanei_usb_read_int. Fixes bug #300878. Added a missing DBG output in sanei_usb_read_int. Check for
usb_interrupt_read to make sure, we have at least version 0.1.8.
* backend/mustek_pp.c, sanei/sanei_pa4s2.c: added support for auto
probing of the port
* configure{.in}, backend/mustek_pp_ccd.*, backend/Makefile.in,
backend/dll.conf, doc/sane-mustek_pp_ccd.man, doc/sane.man,
doc/Makefile.in, doc/descriptions/mustek_pp_ccd.desc: mustek_pp_ccd
backend removed
* backend/mustek_pp.conf, mustek_pp_ccd300.[ch],
doc/sane-mustek_pp.man, doc/descriptions/mustek_pp.desc: fixed CCD
support for mustek_pp backend
* backend/mustek_pp_null.c: beautified debug scanner name
added functions sanei_thread_sendsig() and sanei_thread_get_status()
changed behaviour of sanei_thread_waitpid(), changed parameters
of sanei_thread_begin()
pthread support after all the tests. If libpthread isn't available, try if libc
supports pthreead_*. Use macro USE_PTHREAD to check if threads are used in the
code.
pthread support, added switch --enable-fork-process which toggles
between the usage of fork and pthread, set the default behaviour to
pthread usage on Darwin
* include/sane/config.h: added the HAVE_LIBPTHRAD stuff
* sanei/sanei_thread.c include/sane/sanei_thread.h: fixed compilation issue
on OS/2, added support for fork too, so that we can use sanei_thread
functions in any case.
* backend/plustek.[ch]: changes due to the sanei_thread-lib work
added linker option -lpthread if pthread.h is found
* include/sane/config.in.h: added HAVE_PTHREAD_H
* sanei/sanei_thread.c include/sane/sanei_thread.h: added pthread_
functions to library, so every backend is able to use either fork or
pthread for its reader-process - see plustek backend for example
* backend/plustek.h backend/plustek.c: added sanei_thread stuff to support
threading on MacOS X
* backend/plustek_pp*: some cleanup
* include/sane/sanei_pa4s2.h, sanei/sanei_pa4s2.c: New sanei_pa4s2
version supporting both raw IO and libieee1284. If you don't
have libieee1284 installed, you need to enable raw IO when
running configure
/** Initiate a interrupt transfer read.
*
* Read up to size bytes from the interrupt endpoint from the device to
* buffer. After the read, size contains the number of bytes actually read.
*
* @param dn device number
* @param buffer buffer to store read data in
* @param size size of the data
*
* @return
* - SANE_STATUS_GOOD - on succes
* - SANE_STATUS_EOF - if zero bytes have been read
* - SANE_STATUS_IO_ERROR - if an error occured during the read
* - SANE_STATUS_INVAL - on every other error
*
*/
instead of allocating on stack to avoid crossing 64k borders due to
restrictions of 16-bit device-driver. Cosmetic changes and some casts to
reduce compiler-warnings. Patch from Franz Bakan <fbakan@gmx.net>.
(Linux). Without this fix, especially long running commands and command
queuing would cause memory corruption if the buffer was invalid after the
flush command.