Re: [sane-devel] [janitorial] Code Freeze for 1.0.29 heads up
Hi Allan, m. allan noah writes: > Hmm, this is a fresh install of Fedora 31. I am no longer able to > reproduce the issue, but have done OS updates in the meantime. If you were updating while running ./configure that might explain the "fluke" you reported. Still very unlikely though that you hit a window of timing wherer tiffio.h is installed but libtiff.so not yet. > I'll note that the debian build you linked to does not even have > libusb installed (along with tiff/jpeg, etc), so I wonder how much > code is not being built in that case? Well, according to the logs, still some 77 backends compile and link fine, but I'm not familiar enough with all of them to be able to tell how useful they are. But that wasn't your original point. BTW, in the absence of libusb on Linux the backend will spew debug messages telling you it was built without libusb support. On other platforms (BSDs, BeOS, OS/2, etc.) things may differ. Consult the sanei/sanei_usb.c file for the (very) gory details :-) Back to the hp3900 backend, I looked at the source from a TIFF related build failure point of view and concluded that TIFF is optional as far as that backend is concerned. It is only used by dbg_tiff_save(), for what looks like debugging of calibration, and then only if tiffio.h was found at ./configure time. If not found, dbg_tiff_save() is a noop but the function itself is used unconditionally in the backend's code. In my previous reply I said: >> While I agree that applying this patch is a probably a good idea, there >> is something inconsistent in the checking that I'd like to fix as well. I have now realized that applying the patch is not warranted. All other things equal, the hp3900 backend ought to work just fine in the absence of libtiff # `git grep -i tiff -- backend/hp3900*` is your friend :-) Hope this helps -- Olaf Meeuwissen, LPIC-2FSF Associate Member since 2004-01-27 GnuPG key: F84A2DD9/B3C0 2F47 EA19 64F4 9F13 F43E B8A4 A88A F84A 2DD9 Support Free Softwarehttps://my.fsf.org/donate Join the Free Software Foundation https://my.fsf.org/join
Re: [sane-devel] [janitorial] Code Freeze for 1.0.29 heads up
Hmm, this is a fresh install of Fedora 31. I am no longer able to
reproduce the issue, but have done OS updates in the meantime.
I'll note that the debian build you linked to does not even have
libusb installed (along with tiff/jpeg, etc), so I wonder how much
code is not being built in that case?
allan
On Mon, Jan 27, 2020 at 6:11 AM Olaf Meeuwissen
wrote:
>
> Hi Allan,
>
> m. allan noah writes:
>
> > I noticed that it is not possible to build sane-backends without
> > libtiff installed, because hp3900 backend requires it, but configure
> > does not check for it. This patch seems to fix:
>
> Thanks for noticing and the patch but our CI has been running builds
> without libtiff installed since the very beginning. The latest such
> build for release/1.0.29 is at
>
> https://gitlab.com/sane-project/backends/-/jobs/408050891
>
> Checking the hp3900 source code shows that all TIFF functionality is
> predicated on HAVE_TIFFIO_H. The sane_cv_use_libtiff variable is
> set based on the presence of tiffio.h *and* a TIFFFdOpen function in
> libtiff, though. Would you happen to compile on a system with the
> header file present and no library with TIFFFdOpen by any chance?
>
> # BTW, TIFF seems to be used only by debugging related functionality in
> # the hp3900 backend.
>
> Can you provide the config.log and config.status files for the build
> where you run into this? What distribution/version is this on? What
> are the TIFF related packages that are installed?
>
> > diff --git a/acinclude.m4 b/acinclude.m4
> > index 7c90c64f..f0d7d295 100644
> > --- a/acinclude.m4
> > +++ b/acinclude.m4
> > @@ -598,6 +598,13 @@ for be in ${BACKENDS}; do
> > fi
> > ;;
> >
> > +hp3900)
> > +if test "${sane_cv_use_libtiff}" != "yes"; then
> > + echo "*** $be backend requires TIFF library - $DISABLE_MSG"
> > + backend_supported="no"
> > +fi
> > +;;
> > +
> > dell1600n_net)
> > if test "${sane_cv_use_libjpeg}" != "yes" || test
> > "${sane_cv_use_libtiff}" != "yes"; then
> >echo "*** $be backend requires JPEG and TIFF library - $DISABLE_MSG"
>
> While I agree that applying this patch is a probably a good idea, there
> is something inconsistent in the checking that I'd like to fix as well.
>
> # The TIFF library check was added for the dell1600n_net backend. It
> # may be that the hp3900 backend's link flags picks up a library from
> # one check and activates code based on another check (which happens to
> # be *duplicated* by the library check).
>
> > On Sun, Jan 26, 2020 at 7:02 AM Olaf Meeuwissen
> > wrote:
> >>
> >> Hi all,
> >>
> >> Olaf Meeuwissen writes:
> >>
> >> > Hi all,
> >> >
> >> > Just so you all don't forget ;-)
> >> >
> >> > On 2020-01-26 around 12:00UTC, we'll be entering Code Freeze. This only
> >> > affects the release/1.0.29 branch. The master branch is still open for
> >> > code changes, new features, bug fixes and what not.
> >>
> >> We're past that point in time now and the release/1.0.29 branch is in
> >> Code Freeze. As per milestone[1] description, only "fixes for hardware
> >> destroying bugs, compile failures, fixing completely unusable backends
> >> and documentation fixes are allowed" now.
> >>
> >> [1]: https://gitlab.com/sane-project/backends/-/milestones/3
> >>
> >> [...snip...]
>
> Seeing that this is a compile failure, I'm inclined to include it in the
> upcoming 1.0.29 on the condition that I understand what's going on with
> your build.
>
> Hope this helps,
> --
> Olaf Meeuwissen, LPIC-2FSF Associate Member since 2004-01-27
> GnuPG key: F84A2DD9/B3C0 2F47 EA19 64F4 9F13 F43E B8A4 A88A F84A 2DD9
> Support Free Softwarehttps://my.fsf.org/donate
> Join the Free Software Foundation https://my.fsf.org/join
--
"well, I stand up next to a mountain- and I chop it down with the edge
of my hand"
Re: [sane-devel] [janitorial] Code Freeze for 1.0.29 heads up
Hi Allan,
m. allan noah writes:
> I noticed that it is not possible to build sane-backends without
> libtiff installed, because hp3900 backend requires it, but configure
> does not check for it. This patch seems to fix:
Thanks for noticing and the patch but our CI has been running builds
without libtiff installed since the very beginning. The latest such
build for release/1.0.29 is at
https://gitlab.com/sane-project/backends/-/jobs/408050891
Checking the hp3900 source code shows that all TIFF functionality is
predicated on HAVE_TIFFIO_H. The sane_cv_use_libtiff variable is
set based on the presence of tiffio.h *and* a TIFFFdOpen function in
libtiff, though. Would you happen to compile on a system with the
header file present and no library with TIFFFdOpen by any chance?
# BTW, TIFF seems to be used only by debugging related functionality in
# the hp3900 backend.
Can you provide the config.log and config.status files for the build
where you run into this? What distribution/version is this on? What
are the TIFF related packages that are installed?
> diff --git a/acinclude.m4 b/acinclude.m4
> index 7c90c64f..f0d7d295 100644
> --- a/acinclude.m4
> +++ b/acinclude.m4
> @@ -598,6 +598,13 @@ for be in ${BACKENDS}; do
> fi
> ;;
>
> +hp3900)
> +if test "${sane_cv_use_libtiff}" != "yes"; then
> + echo "*** $be backend requires TIFF library - $DISABLE_MSG"
> + backend_supported="no"
> +fi
> +;;
> +
> dell1600n_net)
> if test "${sane_cv_use_libjpeg}" != "yes" || test
> "${sane_cv_use_libtiff}" != "yes"; then
>echo "*** $be backend requires JPEG and TIFF library - $DISABLE_MSG"
While I agree that applying this patch is a probably a good idea, there
is something inconsistent in the checking that I'd like to fix as well.
# The TIFF library check was added for the dell1600n_net backend. It
# may be that the hp3900 backend's link flags picks up a library from
# one check and activates code based on another check (which happens to
# be *duplicated* by the library check).
> On Sun, Jan 26, 2020 at 7:02 AM Olaf Meeuwissen
> wrote:
>>
>> Hi all,
>>
>> Olaf Meeuwissen writes:
>>
>> > Hi all,
>> >
>> > Just so you all don't forget ;-)
>> >
>> > On 2020-01-26 around 12:00UTC, we'll be entering Code Freeze. This only
>> > affects the release/1.0.29 branch. The master branch is still open for
>> > code changes, new features, bug fixes and what not.
>>
>> We're past that point in time now and the release/1.0.29 branch is in
>> Code Freeze. As per milestone[1] description, only "fixes for hardware
>> destroying bugs, compile failures, fixing completely unusable backends
>> and documentation fixes are allowed" now.
>>
>> [1]: https://gitlab.com/sane-project/backends/-/milestones/3
>>
>> [...snip...]
Seeing that this is a compile failure, I'm inclined to include it in the
upcoming 1.0.29 on the condition that I understand what's going on with
your build.
Hope this helps,
--
Olaf Meeuwissen, LPIC-2FSF Associate Member since 2004-01-27
GnuPG key: F84A2DD9/B3C0 2F47 EA19 64F4 9F13 F43E B8A4 A88A F84A 2DD9
Support Free Softwarehttps://my.fsf.org/donate
Join the Free Software Foundation https://my.fsf.org/join
Re: [sane-devel] [janitorial] Code Freeze for 1.0.29 heads up
Hi, On 2020-01-26 4:02 a.m., Olaf Meeuwissen wrote: Only French has someone stepping up so far. @skelband> Looking forward to an updated po/fr.po :-) I checked in with my daughter yesterday. She is working through it. Hopefully it will be done shortly. Cheers, Ralph
Re: [sane-devel] [janitorial] Code Freeze for 1.0.29 heads up
Hi all, Olaf Meeuwissen writes: > Hi all, > > Just so you all don't forget ;-) > > On 2020-01-26 around 12:00UTC, we'll be entering Code Freeze. This only > affects the release/1.0.29 branch. The master branch is still open for > code changes, new features, bug fixes and what not. We're past that point in time now and the release/1.0.29 branch is in Code Freeze. As per milestone[1] description, only "fixes for hardware destroying bugs, compile failures, fixing completely unusable backends and documentation fixes are allowed" now. [1]: https://gitlab.com/sane-project/backends/-/milestones/3 > Since the Feature Freeze last week, we've seen a couple of translation > updates make their way in. British English, Catalan, German, Ukrainian > and Valencian are all at 100% now! > > There is still one open issue with a merge request that fixes it. It's > nothing spectacular but fixes a compiler warning in the genesys test > suite. That issue (and merge request) have been dealt with since. @povilas> Thanks! Nothing new has popped up, so far, that needs taking care of for 1.0.29 as far as I'm concerned but should you bump into something while testing don't hesitate to submit an issue[2] or bring it up here on the list. [2]: https://gitlab.com/sane-project/backends/issues > Please note that translation updates [are] welcome on release/1.0.29 all > the way up to 2020-02-02 around 10:00UTC and master after that. Feel > like helping out, check the list with translation issues at > > https://gitlab.com/sane-project/backends/issues?label_name%5B%5D=translation The above re translation updates still stands. Submissions are welcome as merge requests (preferred) as well as patches. > Only French has someone stepping up so far. @skelband> Looking forward to an updated po/fr.po :-) # Totally unrelated to the Code Freeze, I also cleaned up the artifacts # of our old (pre 2019-09-01) CI pipelines. This removed some 230 of # the source tarballs that were still lingering around, roughly halving # their number. At present, snapshot tarballs older than 90 days are # nuked. Build logs are kept and releases are uploaded elsewhere[3] as # a copy to preserve them for eternity/posterity. Just FYI. # # [3]: https://gitlab.com/sane-project/backends/-/releases Hope this helps, -- Olaf Meeuwissen, LPIC-2FSF Associate Member since 2004-01-27 GnuPG key: F84A2DD9/B3C0 2F47 EA19 64F4 9F13 F43E B8A4 A88A F84A 2DD9 Support Free Softwarehttps://my.fsf.org/donate Join the Free Software Foundation https://my.fsf.org/join
