This driver is neither working nor has it been in a compiling state for
a long time, so unhook it from the build until it is fixed and works.
The files (api.c and protocol.[ch]) are still in git, but won't end up in
released tarballs and they don't get built (neither git nor tarballs).
This also allows us to drop the otherwise unneeded dependency on libudev.
When the MSO-19 driver comes back, it should be in a form that doesn't
require the inherently Linux-only libudev anyway. See also:
http://sigrok.org/bugzilla/show_bug.cgi?id=65
This driver has been unmaintained for years, and was never good code
to begin with. It's also questionable whether it was ever useful,
particularly with the demo driver now supporting various analog
signalling.
Instead of >= 44 Makefile.am's we now only have one top-level
Makefile.am, and use the 'subdir-objects' automake option to
handle the build via non-recursive (auto)make.
This has the advantage of fewer (boilerplate or other) files and less
clutter in general, as well as performance advantages since the new
setup can build many files in parallel (with 'make -j'), not only 2 or 3
files within the same (e.g. hardware/xxxx/* subdirectory) and also since
we no longer need to build intermediate libtool helper libs per subdirectory.
A quick, non-scientific test build on a quad-core laptop with 'make -j 4'
yields a build time reduction from 35s to 19s.
All autotools features that worked before are still intact without any
regressions, including the Make targets 'install', 'uninstall', 'check',
'dist', 'clean', 'distclean' and so on, as well as all the usual portability
handling (build works on any OS, with any Make implementation such as
GNU Make or BSD Make, with any shell such as sh/ksh/zsh/bash/dash, etc. etc.)
and features such as out-of-tree build support, cross-compile support,
testsuite support (also with colored output), "silent make rules", etc. etc.
This gets us the libusb version checking mechanism itself, hopefully
making this sort of thing easier in future.
Also hotplug, device tree traversal, and lots of fixes.
Disable drivers that need serial port support if libserialport is not found.
Also, disable building various other serial port related code in that case.
The libtool current:revision:age numbers change from 1:1:0 to 1:2:0
(i.e., revision is increased) since the library source code has changed,
but no interfaces were added or changed or removed.
Details:
http://www.gnu.org/software/libtool/manual/libtool.html#Updating-version-info
This changes the library filename (e.g. on Linux) from libsigrok.so.1.0.1
to libsigrok.so.1.0.2, but the SONAME (+symlink) remains the same
(libsigrok.so.1) since this release is API- and ABI-compatible with the last.
We now use zip_int64_t and zip_get_num_entries() for example, which
requires at least libzip 0.10. This version was released in 03/2012,
which is old enough that we don't necessarily have to do a work-around
for older versions. Thus, simply bump the requirement to >= 0.10.
The last release (0.2.0) had the libtool version (current:revision:age)
set to 1:0:0. Since this release doesn't add/change/remove any
interfaces, only 'revision' is increased, resulting in 1:1:0.
http://www.gnu.org/software/libtool/manual/libtool.html#Updating-version-info
Frontends using libsigrok don't need to be recompiled or relinked.
These two drivers are currently unfinished and don't work, so disable
and "unhook" them for now in preparation of the next libsigrok release.
They're still in the git repository, but not hooked up to the build
system, so that they won't get detected or built, and also don't end up
in the release tarball.
Since link-mso19 is the only driver that currently requires libudev,
drop any reference to that, too.
It should be relatively easy to apply this patch in reverse after the
release to bring back both drivers.
These two drivers are currently unfinished and don't work, so disable
and "unhook" them for now in preparation of the next libsigrok release.
They're still in the git repository, but not hooked up to the build
system, so that they won't get detected or built, and also don't end up
in the release tarball.
Since link-mso19 is the only driver that currently requires libudev,
drop any reference to that, too.
It should be relatively easy to apply this patch in reverse after the
release to bring back both drivers.
When checking architecture-specific things, always check $host, i.e. the
architecture we're building _for_, not the one we happen to build _on_.
E.g. when cross-compiling _for_ Android (or Windows or others) it's important
to check for Android in $host; whether we happen to cross-compile _on_ a Linux
or Windows or OpenBSD or FreeBSD machine ($build) doesn't matter, only the
fact that we compile _for_ Android is important for most checks.
In the configure summary at the end also print the architecture we're
building on ($build) and the target host we build for ($host). The two are
not necessarily the same, e.g. in the case of cross-compiles.
In the summary output at the end of a configure run, explicitly mention
which versions of which libraries are required, and also the version which
pkg-config has found.
Use the canonical driver name (all-lowercase, e.g. "serial-dmm") in the
list of enabled/disabled drivers that configure prints after a run.
It's common to many drivers that they support multiple devices, so
printing one device name (e.g. "ChronoVu LA8") is seldom really correct.
E.g. the agilent-dmm, asix-sigma, brymen-dmm, colead-slm, fluke-dmm,
fx2lafw, hantek-dso, lascar-el-usb, mic-985xx, openbench-logic-sniffer,
rigol-ds1xx2, uni-t-dmm, victor-dmm, and zeroplus-logic-cube drivers
all support more than just one device.
So, just print the driver name instead which is more correct anyway
since it's specifically a list of enabled/disabled drivers.
Don't rely on the "heuristic" that 'libusb_CFLAGS' will be non-empty if
libusb-1.0 was found, but rather use the proper method of checking the
variable 'have_libusb1_0' which pkg-config will set to "yes"/"no"
depending on whether it finds the library.
- If libusb-1.0 is not found, do not compile in ezusb.c and usb.c since
they require libusb.h. The respective hardware drivers that use
libusb-1.0, and usb.c and/or ezusb.c will be excluded from the build
elsewhere in configure.ac. Rename NEED_EZUSB to NEED_USB.
- Drop the NEED_SERIAL check and always compile in serial.c. This is a
very small chunk of code, it does not depend on any external
libraries that might be missing, and it compiles on all architectures.
Thus there's no need to conditionally include or exclude it.
In order for lib autodetection and disabling of drivers (which require
libs that cannot be found) to work, the order of AC_ARG_ENABLEs,
AM_CONDITIONALs, and AC_DEFINEs needs to be changed.
All drivers start out enabled or disabled (depending on whether
the --disable-drivers option was supplied or not). Then, any driver can
be enabled or disabled explicitly via --enable-<drivername> or the
resp. --disable-<drivername> option. Finally, pkg-config checks for all
libraries are performed, and all drivers which require a library that
cannot be found are disabled explicitly (regardless of any
--enable-<drivername> option that might have been supplied).
Until now, we checked for certain (optional) libraries via pkg-config and
the configure script would abort if any of them was not found, even
though they were optional. It was up to the user to then figure out which
combination of --disable-<drivername> switches were required for his
specific OS (and set of installed libs) to get a working configure run.
Only if the user already specified enough --disable-<drivername>
switches beforehand, so that all drivers which require a missing library
were disabled, would the configure run not check for that specific lib
(and would thus not fail).
With this change, we now always unconditionally check for all libs
(required and optional) via pkg-config. However, whether an (optional) lib
is found or not, configure will not abort. Instead, it'll just disable
all drivers which need a lib that cannot be found.
The user will no longer have to supply --disable-<drivername> parameters
in order to get a working build.
The automake 'std-options' option checks whether all installed tools
and scripts have a --help and --version CLI option. This check is not
needed for libsigrok though, since it doesn't install any tools.
This adds a suite of unit tests for libsigrok. It uses the 'Check'
tool/library (apt-get install check) to run the tests.
The configure tool tries to find libcheck. If it succeeds, a
"make check" will run all tests. Otherwise, none of the tests will
be built and "make check" will not run any tests.
This also means that users who don't have 'check' installed will still
be able to build and install libsigrok just fine.
We don't need a C++ compiler for libsigrok, so don't check for one.
This would cause a build failure on all systems where there is a C compiler
but no C++ compiler installed.
We no longer use gthread (its only user was the 'demo' driver), so drop
the 'configure' check for it.
The libsigrok.pc pkg-config file is still correct, since the 'configure'
result propagates there automatically.
Don't override/overwrite CFLAGS in configure.ac, but rather amend it
with (currently) "-Wall -Wextra -fvisibility=hidden".
This properly allows users/packagers to do things like:
./configure (this will default to using "-g -O2" additionally)
CFLAGS="-g -O2" ./configure (same as above)
CFLAGS="" ./configure (no additional flags)
CFLAGS="-g -O0" ./configure (disable optimization, e.g. for valgrind use)
etc. etc.
Uwe Hermann
Martin Ling
Bert Vermeulen
Aurelien Jacobs
Matthias Heidbrink
Daniel Elstner
Marc Schink
poljar (Damir Jelić)
Dan Horák
Matthias Heidbrink
Marcus Comstedt
Alexandru Gagniuc