Re: [gudjon@gudjon.org: Re: [comedi] libcomedi0 removed from debian testing]
Yaroslav Halchenko
have slipped through my mailbox without notice ;)
Gudjon, I would be glad (whenever time allows) to sponsor the
upload to Debian. I will look into current package you have -- only 1
major issue I see:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=570069
if it is usable on kfreebsd then udev depends should be said
[!kfreebsd-amd64] or smth like that OR may be moved into Recommends
altogether, OR just do not build for kfreebsd if it is of no use there
-- I am just ignorant on this subject. Any input is welcome to help
resolving this.
GIT version -- would be neat (whenever current one gets back into
testing)!
On Thu, 05 Apr 2012, Christoph Schmidt-Hieber wrote:
> Hi Yaro,
> maybe NeuroDebian can come to the rescue here? Comedi is an incredibly
> important package, and it would be a shame to see it vanish from the
> debian repositories.
> Cheers,
> Christoph
> ----- Forwarded message from "Gudjon I. Gudjonsson" <gud...@gudjon.org> -----
> Date: Thu, 5 Apr 2012 14:12:58 +0100
> From: "Gudjon I. Gudjonsson" <gud...@gudjon.org>
> To: "comed...@googlegroups.com" <comed...@googlegroups.com>
> Subject: Re: [comedi] libcomedi0 removed from debian testing
> user-agent: KMail/1.13.7 (Linux/3.1.0-1-686-pae; KDE/4.6.5; i686; ; )
> Hi list
> There is a newer version on my private server (0.8.1-6)
> deb-src http://213.64.124.152/debian/ source/
> but I haven't had time to test it properly.
> My plan was to upload version 0.8.1-6 wich includes symbols file and then make
> a package with the current git version.
> My problem is lack of time and that I am not a Debian Maintainer. If someone
> wants to take over the maintenance, please do. If somebody is willing to
> sponsor uploads for me, I am willing to keep on maintaining it but I have been
> busy lately and will be for 10 more days.
> Cheers
> Gudjon
> On Thursday 05 April 2012 14:52:53 Bernd Porr wrote:
> > Seems so that it needs a new maintainer. I vaguely remember that
> > somebody here wanted to volunteer and that I offered testing the
> > package. Anybody?
> > /Bernd
> > Daniele Nicolodi wrote:
> > > On 05/04/12 14:16, Bernd Porr wrote:
> > >> Hi all,
> > >> does anybody know why libcomedi0 has been removed from debian testing?
> > > It looks like this RC bug is the reason:
> > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=570069
> > > This bug is more than two years old, and the comedilib Debian package
> > > has not seen any activity since 2008, except a NMU in 2011...
> > > http://packages.qa.debian.org/c/comedilib.html
> > > Cheers,
--
=------------------------------------------------------------------=
Keep in touch www.onerussian.com
Yaroslav Halchenko www.ohloh.net/accounts/yarikoptic
Bernd Porr
not sure if you have seen it on the mailing list. Graeme has reported a
bug in the new libcomedi:
------------
I have recently installed the libcomedi0 package from Debian unstable
(my distro is Linux Mint Debian Edition) and found a couple of
problems.
Firstly I couldn't access the comedi device without sudo rights, even
after adding myself to the iocard group. It seems that the
comedi.rules file must be moved to the /udev/rules.d directory for
this to work.
------------
Ironically UBUNTU also updated it's libcomedi yesterday but the version
0.8.1-5 and not 0.8.1-6. The iocard/udev stuff still works under UBUNTU
precise but no longer under Mint.
/Bernd
--
www: http://www.berndporr.me.uk/
http://www.linux-usb-daq.co.uk/
http://www.imdb.com/name/nm3293421/
Mobile: +44 (0)7840 340069
Work: +44 (0)141 330 5237
University of Glasgow
School of Engineering
72 Oakfield Avenue (Rankine Building for deliveries)
Glasgow, G12 8LT
Bernd Porr
the command:
./configure --with-udev-hotplug=/lib
should now place all udev related scripts in /lib and none in /etc. That
was a mistake in the configure.ac (sysconf=/etc instead of the argument
of --with-udev-hotplug).
http://lintian.debian.org/tags/udev-rule-in-etc.html
/Bernd
Bernd Porr
I've added an option to disable the firmware installation especially for
debian/Ubuntu package maintainers.
./configure --help
--disable-firmware Disable installation of firmware files
Bernd Porr
I've just done a fresh install of the debian testing and the usbdux
firmware files comes with the package "firmware-linux-free". So, to
avoid having them included twice in the firmware package and in
libcomedi can you disable them with in libcomedi with --disable-firmware
in the ./configure script.
/Bernd
Glasgow, G12 8LT
Bernd Porr
I've just tried to build a new debian package for libcomedi based on the
latest git package Gudjon has provided. It fails trying to patch the
comedi-calibrate. It cannot find the ./configure script (though it's there).
Here's everything: the original package and my attempt to create a new
version.
http://www.linux-usb-daq.co.uk/tmp/comedilib.tar.gz
Anyway. I've corrected the ./configure call in rules using the new
options we have added during the last couple of days. Also played a bit
with the build dependencies (removed ruby there, added dblatex, ...).
/Bernd
http://www.berndporr.me.uk/
Mobile: +44 (0)7840 340069
Work: +44 (0)141 330 5237
University of Glasgow
School of Engineering
Ian Abbott
> Hi,
>
> I've just tried to build a new debian package for libcomedi based on the
> latest git package Gudjon has provided. It fails trying to patch the
> comedi-calibrate. It cannot find the ./configure script (though it's there).
>
> Here's everything: the original package and my attempt to create a new
> version.
> http://www.linux-usb-daq.co.uk/tmp/comedilib.tar.gz
>
> Anyway. I've corrected the ./configure call in rules using the new
> options we have added during the last couple of days. Also played a bit
> with the build dependencies (removed ruby there, added dblatex, ...).
removed from the release (non-git) version of the comedilib source
packages (once we finally get round to releasing the thing!).
--
-=( Ian Abbott @ MEV Ltd. E-mail: <abb...@mev.co.uk> )=-
-=( Tel: +44 (0)161 477 1898 FAX: +44 (0)161 718 3587 )=-
Gudjon I. Gudjonsson
Hi
On Thursday 03 May 2012 11:37:43 Bernd Porr wrote:
> I've just tried to build a new debian package for libcomedi based on the
> latest git package Gudjon has provided. It fails trying to patch the
> comedi-calibrate. It cannot find the ./configure script (though it's
> there).
It is the comedi_calibrate configure script. Did you create your package as a multiple source package?
I am going on a holiday for a few days but I will se if I can upgrade my package to the latest git.
>
> Here's everything: the original package and my attempt to create a new
> version.
> http://www.linux-usb-daq.co.uk/tmp/comedilib.tar.gz
>
> Anyway. I've corrected the ./configure call in rules using the new
> options we have added during the last couple of days. Also played a bit
> with the build dependencies (removed ruby there, added dblatex, ...).
Do you like the multiple source approach?
Why is comedi_calibrate a separate source?
Can you please add some command line parameter that makes configure not search for libcomedi.so while configuring comedi_calibrate? Then I don't have to patch configure. Patching configure is a bad habit since the configure scripts can be regenerated and then the patch doesn't work anymore.
There are dependencies on unnecessary libraries. Is it possible/easy to fix that before the release?
Regards
Gudjon
Bernd Porr
data again. Binary wise: comedi_board_info. Ian (and the others here)
could you test that with a couple of other boards? I'm obviously a bit
biased towards the USB-DUX... (buy them!).
/Bernd
Ian Abbott wrote:
> On 2012/05/03 10:37 AM, Bernd Porr wrote:
>> Hi,
>>
>> I've just tried to build a new debian package for libcomedi based on the
>> latest git package Gudjon has provided. It fails trying to patch the
>> comedi-calibrate. It cannot find the ./configure script (though it's there).
>>
>> Here's everything: the original package and my attempt to create a new
>> version.
>> http://www.linux-usb-daq.co.uk/tmp/comedilib.tar.gz
>>
>> Anyway. I've corrected the ./configure call in rules using the new
>> options we have added during the last couple of days. Also played a bit
>> with the build dependencies (removed ruby there, added dblatex, ...).
>
> Of course the dblatex and other documentation build dependencies can be
> removed from the release (non-git) version of the comedilib source
> packages (once we finally get round to releasing the thing!).
>
--
Bernd Porr
> search for libcomedi.so while configuring comedi_calibrate?
script. It's in the comedi_calibrate git now.
/Bernd
> You received this message because you are subscribed to the Google
> Groups "Comedi: Linux Control and Measurement Device Interface" group.
> To post to this group, send email to comed...@googlegroups.com.
> To unsubscribe from this group, send email to
> comedi_list...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/comedi_list?hl=en.
Ian Abbott
> > Can you please add some command line parameter that makes configure not
> > search for libcomedi.so while configuring comedi_calibrate?
>
> Just added the option "--disable-comedilib-checking" to the configure
> script. It's in the comedi_calibrate git now.
Wouldn't the comedi_calibrate build fail during the linking stage if
libcomedi.so is not there?
Bernd Porr
scripts run and then the two makes are run later. To get this to compile
also the comedilib header check needs to be disabled. Basically because:
comedilib/configure
comedilib/comedi-calibrate/configure
comedilib/make
comedilib/comedi-calibrate/make
comedilib/make install
comedilib/comedi-calibrate/make install
I'll hand this now over to Gudjon who has more experience with this:
http://www.linux-usb-daq.co.uk/tmp/comedilib.tar.gz
in there is his original tar and my hacked one. There's still something
not right with the firmware / hotplug stuff. The configure seems to be
ignoring all the options.
/Bernd
Ian Abbott wrote:
> On 2012/05/03 11:43 AM, Bernd Porr wrote:
>> > Can you please add some command line parameter that makes configure not
>> > search for libcomedi.so while configuring comedi_calibrate?
>>
>> Just added the option "--disable-comedilib-checking" to the configure
>> script. It's in the comedi_calibrate git now.
>
> I'm curious to know why this is required; it seems rather odd to me!
> Wouldn't the comedi_calibrate build fail during the linking stage if
> libcomedi.so is not there?
>
--
Ian Abbott
merging the comedi_calibrate sources back in to the comedilib sources.
If comedi_calibrate had its own debian source package this extra
configure option wouldn't be needed.
Might be better off avoiding needless extra comedi_calibrate configure
options and either splitting the debian source package or retro-patching
the comedi_calibrate stuff back in at the autoconf/automake level,
basically undoing these two patches:
<http://comedi.org/git?p=comedi/comedilib.git;a=blobdiff;f=configure.ac;h=bcc1ae532bd97c5cdf5444e8923e1c33763d6f90;hp=4740a18a1ac4bba20ec80e04c98823342b8e921b;hb=c632d4e636aa89acd4543aea4d04aa32ac05da94;hpb=d4fd5f7b087d96676f53d3d74fde681baa26270c>
<http://comedi.org/git?p=comedi/comedilib.git;a=blobdiff;f=Makefile.am;h=6abfe2227c32553f416bb13382bba451fb01e10e;hp=8a75b7b587bf413d17477bd9f4cc1c8563cded85;hb=64d90d7f374c64cdce46aff0e397ba35f33563e6;hpb=c632d4e636aa89acd4543aea4d04aa32ac05da94>
Bernd Porr
some dependency problems? We could def get comedi_calibrate back in with
the option to enable/disable it. Would make life much easier.
/Bernd
Mobile: +44 (0)7840 340069
Work: +44 (0)141 330 5237
University of Glasgow
School of Engineering
Ian Abbott
> Do you remember why we separated them in the first place? Were there
> some dependency problems? We could def get comedi_calibrate back in with
> the option to enable/disable it. Would make life much easier.
> /Bernd
dependencies that the rest of comedilib doesn't have.
It was taken out back in May 2006 - that's nearly a year before
comedilib-0.8.0 was released.
Bernd Porr
the reason why comedilib was quite stable in debian for years was the
few dependencies.
I guess in an ideal world we would like to have two different packages:
comedilib and comedi-calibrate. Gudjon, I had a look in the debian bug
system and googled a bit around. One person submitted a bug report and
requested a package for comedi calibrate in 2009 but nobody else ever
moaned about it. There seem to be only a very few people using comedi
calibrate. I think we could just leave comedi calibrate in the git
repository as it is. If Guido has at some point has massive spare time
he could then do the package whenever it's convenient for him. How about
that? But that the priority is with comedilib for now.
Mobile: +44 (0)7840 340069
Work: +44 (0)141 330 5237
University of Glasgow
School of Engineering
Apodaca, Bob
> One person submitted a bug report and requested a package for comedi
> calibrate in 2009 but nobody else ever moaned about it. There seem to
> be only a very few people using comedi calibrate. I think we could
> just leave comedi calibrate in the git repository as it is.
I'm not 100% certain anymore.
Maybe I do not understand the purpose of the comedi calibration
program. My assumption was to reduce/eliminate biases based on
operating temperatures, etc.
Bernd Porr
just add these corrections to my applications because usually there are
all different kinds of sensors connected where calibration can involve
anything from applying a voltage to boiling water and then dropping a
temperature sensor in it.
/Bernd
Bernd Porr
revert).
How do we move forward?
I think comedi-calibrate should stay a separate package and for now as a
source package as for the last couple of years. Ideally it also requires
a person who collects calibration data and writes new code for newer
cards which extracts the data who then should be also the maintainer of
the package (volunteers?).
Ian is right that comedi-calibrate depends on quite a lot of other
compile and runtime packages and will probably break much quicker its
dependencies.
Let's get this release of comedilib done so that we can build a new
package soon. The main issue was the documentation and that's now in a
pretty good shape. BTW on debian testing it needs to be fop and not dblatex.
What do the others think?
/Bernd
Ian Abbott
> I've removed the option from configure in comedi calibrate again (git
> revert).
>
> How do we move forward?
>
> I think comedi-calibrate should stay a separate package and for now as a
> source package as for the last couple of years. Ideally it also requires
> a person who collects calibration data and writes new code for newer
> cards which extracts the data who then should be also the maintainer of
> the package (volunteers?).
comedi_configure could continue more easily without updating the
comedilib package each time.
> Ian is right that comedi-calibrate depends on quite a lot of other
> compile and runtime packages and will probably break much quicker its
> dependencies.
>
> Let's get this release of comedilib done so that we can build a new
> package soon. The main issue was the documentation and that's now in a
> pretty good shape. BTW on debian testing it needs to be fop and not dblatex.
dblatex output. I want to try passing an XSL stylesheet to it.
Yaroslav Halchenko
On Fri, 04 May 2012, Bernd Porr wrote:
> Ian is right that comedi-calibrate depends on quite a lot of other
> compile and runtime packages and will probably break much quicker
> its dependencies.
number of dependencies a burden to maintain across different releases,
otherwise number of dependencies is a no issue in the Debian world ;) I
maintain quite a few packages which pull in hundreds of
dependencies to build, but with proper specification/compatibility layer
I rarely run into any problems due to "large number of dependencies".
Gudjon I. Gudjonsson
Hi
Now I have upgraded the package on my server to 0.8.2~git20120507-1.
On Thursday 03 May 2012 19:08:37 Bernd Porr wrote:
> Good point. I've just struggled with these build dependencies. I guess
> the reason why comedilib was quite stable in debian for years was the
> few dependencies.
I reverted some of the changes to the dependencies you had made.
With the current dependencies it builds under pbduilder.
>
> I guess in an ideal world we would like to have two different packages:
> comedilib and comedi-calibrate. Gudjon, I had a look in the debian bug
> system and googled a bit around. One person submitted a bug report and
> requested a package for comedi calibrate in 2009 but nobody else ever
> moaned about it. There seem to be only a very few people using comedi
> calibrate. I think we could just leave comedi calibrate in the git
> repository as it is. If Guido has at some point has massive spare time
> he could then do the package whenever it's convenient for him. How about
> that? But that the priority is with comedilib for now.
I agree, if nobody uses it then we can ignore the bug. But with my ugly patch it builds and works nicely as a multiple source package.
Linking comedi_calibrate doesn't seem to be neccessary. I can build comedilib and comedi_calibrate without having /usr/lib/libcomedi.so installed.
It is not more work for me to include comedi_calibrate but I will remove it if you really want to.
I will try to remove the unneeded extra libraries myself from the build system and send you a patch. It would be nice to have that fixed before release.
Regards
Gudjon
Yaroslav Halchenko
I would be glad to mentor/sponsor libcomedi upload to Debian sid as soon
as things finalize (to ship or not calibrate) -- Debian wheezy freeze is
coming (in June) so it would be great to get recent libcomedi into
wheezy/sid asap so it gets tested by people before the release.
Cheers!
Yaroslav O. Halchenko
Postdoctoral Fellow, Department of Psychological and Brain Sciences
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834 Fax: +1 (603) 646-1419
WWW: http://www.linkedin.com/in/yarik
Gudjon I. Gudjonsson
Hi
On Monday 07 May 2012 14:43:14 Yaroslav Halchenko wrote:
> since time is running out I will chime in with possibly useless:
> I would be glad to mentor/sponsor libcomedi upload to Debian sid as soon
> as things finalize (to ship or not calibrate) -- Debian wheezy freeze is
> coming (in June) so it would be great to get recent libcomedi into
> wheezy/sid asap so it gets tested by people before the release.
comedi_calibrate isn't a problem for a Debian upload.
I find the shlibdeps warnings problematic and would like to have them fixed before upload but it is not neccessary.
Shall we make an upload to Debian before or after comedi release? After is of course preferable.
Cheers
Gudjon
Bernd Porr
know Ian is still tinkering with the formatting of the documentation but
I think we are very close to a release (we just need to wait until he's
getting tired messing in the style sheets ;). Ian?
I'm also happy if comedi-calibrate is included (not sure if anybody is
really using it but if it's not a hassle then yes of course include it).
/Bernd
Gudjon I. Gudjonsson wrote:
> You received this message because you are subscribed to the Google
> Groups "Comedi: Linux Control and Measurement Device Interface" group.
> To post to this group, send email to comed...@googlegroups.com.
> To unsubscribe from this group, send email to
> comedi_list...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/comedi_list?hl=en.
Yaroslav Halchenko
On Mon, 07 May 2012, Gudjon I. Gudjonsson wrote:
> of course preferable.
testing its build across archs by uploading first pre-release of
experimental
Ian Abbott
> From my point of view libcomedi can be released as soon as possible. I
> know Ian is still tinkering with the formatting of the documentation but
> I think we are very close to a release (we just need to wait until he's
> getting tired messing in the style sheets ;). Ian?
remaining with the PDF produced by dblatex:
1. Links to anchors within program listings appear as hyperlinks within
the PDF, but they go to the top of the document instead of the anchor
position. Links within program listings are not converted to hyperlinks
in the PDF.
At least dblatex is nice enough to warn you that it's ignoring link and
anchor tags (and most other tags) in the programlistings:
link: default template used in programlisting or screen
anchor: default template used in programlisting or screen
(lots of those...)
2. Rendering of <funcsynopsis> function prototypes is a bit ugly
compared to the HTML output. The HTML output puts one parameter per
line, indented like this:
double comedi_to_phys( lsampl_t data,
comedi_range *range,
lsampl_t maxdata);
but the PDF output puts them all in a single (wrapped) line, like this:
double comedi_to_phys(lsampl_t data, comedi_range * range, lsampl_t
maxdata);
3. Hyphenation of function names etc. is a bit hap-hazard. Basically it
just hyphenates wherever it likes. Still, that's better than text
running into the right-hand margin (and off the page!) like it was doing
earlier.
I think we could probably live with those problems, and I'm tired with
messing with the docs and style-sheets!
Bernd Porr
I think one thing should be changed:
the functions "comedi_to_phys" and "phys_to_comedi" should go back to
the normal function reference. Polynomials are a nice add on (and
requiring comedi_calibrate, special code, ...) but for 99% of all cases
the standard linear mapping between raw data and physical values is fine
and works with every card. Personally I've never bothered with the
polynomials because non-linearlities come mainly from external sources
such as sensor non-linearity or amplifier drift etc and that usually
non-linearity of the ADC is negligible. Because of this the calibrate
functions would need to integrate all these externals factors. A sigma
delta converter (as in the DUX sigma) is terribly linear whereas a
thermocouple connected to it needs to be corrected by polynomials. The
comedi_to_phys is straightforward to use and it's instantly clear how to
plug in the range structures and the maxdata into it. Shall we move them
back to the main doc? I would love to have them back there (I never used
comedi_to_physical in the many years I've been using comedi).
/Bernd
Ian Abbott
> Ian,
>
> I think one thing should be changed:
>
> the functions "comedi_to_phys" and "phys_to_comedi" should go back to
> the normal function reference.
Bernd Porr
/Bernd
Ian Abbott wrote:
> On 2012-05-10 20:27, Bernd Porr wrote:
>> Ian,
>>
>> I think one thing should be changed:
>>
>> the functions "comedi_to_phys" and "phys_to_comedi" should go back to
>> the normal function reference.
>
> I agree. I'll fiddle around with the document a bit more!
>
--