Migrate DEB packages to antimicrox

[gombosg]

The DEB package was packaged last year. If possible, migrate users to antimicrox as that is the maintained version.

Goal:

• contact the Debian package maintainer, if possible continue an open discussion here
• learn about DEB requirements

[gombosg]

I'm happy to start the conversation, @pktiuk do you have any specific inputs/suggestions before I contact the maintainer?

[pktiuk]

Just tell ask him whether does he use github.
If yes, he could answer us here.

[gombosg]

We had an email conversation. The Debian packager will package antimicrox for Debian, replacing antimicro. For this we need config migration ASAP since Debian stable freeze is in about a month.

I'll look into config migration this week, if the formats themselves are compatible it shouldn't be an issue.

[mirabilos]

I’ve begun packaging 3.1.3 and have a few comments already:

• list(APPEND CMAKE_PREFIX_PATH "/usr/lib/arm-linux-gnueabihf") is a complete no-go; the packager/builder should be expected to set this correctly. Doing that in the CMakeLists file will cause build errors if building for some platform while having the armhf variants installed.

• please add an option to not build a shared library (i.e. link the executable with the static library and not install the library); otherwise, I’d have to split the package and you’d have to commit to a stable ABI, which I’m sure you don’t want

• hardcoding LD_LIBRARY_PATH in the binary is, similarily, a no-go, but I assume that would be a nōn-issue if the shared library is disabled

• there’s a CHANGELOG.md and a HISTORY.md, can you decide on one of them? They both seem to be changelog-like files, not short NEWS-like ones.

• files have the Unix executable attribute that shouldn’t (such as LGPL licence files, source/UI files, …); has someone committed from Windows there? ;-)

• some even have a Byte-Order Mark that shan’t be there

• the xbox fakeclasses and all tests/ files don’t have explicit copyright headers (not a problem, just an inconsistency)

I’ll probably find more and report.

[mirabilos]

• qglobalshortcut is not licenced (showstopper)

Reported as: mitei/qglobalshortcut#1

[AriaMoradi]

• qglobalshortcut is not licenced (showstopper)

Reported as: mitei/qglobalshortcut#1

The original dev doesn't have any activity record for the past 5 years, hope they are actually around.

[gombosg]

Thanks @mirabilos!

I created issues and a project to track progress: https://github.com/AntiMicroX/antimicrox/projects/1

@AntiMicroX/members feel free to take issues.
@AriaMoradi #99 becomes high priority. We have to fix the pipeline somehow.

[pktiuk]

@mirabilos
Do I correctly assume we should also deal with all the problems reported by lintian for our package?
Current status

E: antimicrox: no-copyright-file
W: antimicrox: desktop-mime-but-no-exec-code usr/share/applications/io.github.antimicrox.antimicrox.desktop
W: antimicrox: extended-description-line-too-long
W: antimicrox: non-standard-dir-perm usr/ 0775 != 0755
W: antimicrox: non-standard-dir-perm usr/bin/ 0775 != 0755
W: antimicrox: non-standard-dir-perm usr/include/ 0775 != 0755
W: antimicrox: non-standard-dir-perm ... use --no-tag-display-limit to see all (or pipe to a file/program)
W: antimicrox: syntax-error-in-debian-changelog line 10 "badly formatted heading line"
W: antimicrox: syntax-error-in-debian-changelog line 102 "badly formatted heading line"
W: antimicrox: syntax-error-in-debian-changelog line 108 "badly formatted heading line"
W: antimicrox: syntax-error-in-debian-changelog ... use --no-tag-display-limit to see all (or pipe to a file/program)
W: antimicrox: wrong-name-for-changelog-of-native-package usr/share/doc/antimicrox/changelog.Debian.gz

(but there is already PR with some fixes for this #117)

[mirabilos]

Paweł Kotiuk dixit:

Do I correctly assume we should also deal with all the problems reported by `lintian` for our package?

No, because I’m not going to use anything from your package.

Current status

Looking at that it’d be harder to fix your package than to make one anew, which I don’t even need because I already have antimicro packaging that I can reuse. In fact, it is preferred if upstreams don’t ship any packaging, and Debian’s tooling will automatically remove a debian/ subdir if it exists, as we ship our own packaging.

[gombosg]

@mirabilos I think we have fixed the issues you mentioned.
You can check the items at https://github.com/AntiMicroX/antimicrox/projects/1.

Please do a test build on master f27aa8d if possible, and if you find it acceptable we'll make a release for the final package.

[pktiuk]

@mirabilos
Do you have any status updates for us?
We are thinking about pushing a new release soon, is it okay with your packaging efforts?

[mirabilos]

Paweł Kotiuk dixit:

@mirabilos Do you have any status updates for us?

Sorry, haven’t yet had any spare time.

We are thinking about pushing a new release soon, is it okay with your packaging efforts?

Just do, I’ll update the packaging as needed. In fact having things in releases helps. bye, //mirabilos -- <igli> exceptions: a truly awful implementation of quite a nice idea. <igli> just about the worst way you could do something like that, afaic. <igli> it's like anti-design. <mirabilos> that too… may I quote you on that? <igli> sure, tho i doubt anyone will listen ;)

[mirabilos]

CHANGELOG.md is installed during build:

-- Installing: /tmp/buildd/antimicro-3.1.4/debian/antimicro/usr/share/doc/antimicrox/CHANGELOG.md
-- Installing: /tmp/buildd/antimicro-3.1.4/debian/antimicro/usr/share/antimicrox/CHANGELOG.md

Twice is bad, once would be better, but since I install it in the correct Debian location using Debian mechanisms anyway it doesn’t need to be installed at all (not for Debian anyway), but I can just delete the file after make install.

Given that src/aboutdialog.cpp loads it from an embedded resource, I see no reason for cmake/make to install that file at all.

[mirabilos]

The headers are also still installed; probably a remnant of the now-removed library. Since no library is installed, nothing can really make use of these headers, so they need not be installed.

[mirabilos]

Compiling goes well; there’s one file with warnings:

[ 98%] Building CXX object CMakeFiles/antimicrox.dir/src/eventhandlers/uinputeventhandler.cpp.o
/usr/bin/c++ -DQT_CONCURRENT_LIB -DQT_CORE_LIB -DQT_GUI_LIB -DQT_NETWORK_LIB -DQT_NO_DEBUG -DQT_WIDGETS_LIB -DWITH_UINPUT -DWITH_X11 -DWITH_XTEST -I/tmp/buildd/antimicro-3.1.4/obj-x86_64-linux-gnu -I/tmp/buildd/antimicro-3.1.4 -I/tmp/buildd/antimicro-3.1.4/obj-x86_64-linux-gnu/src -I/tmp/buildd/antimicro-3.1.4/src -isystem /usr/include/x86_64-linux-gnu/qt5 -isystem /usr/include/x86_64-linux-gnu/qt5/QtCore -isystem /usr/lib/x86_64-linux-gnu/qt5/mkspecs/linux-g++ -isystem /usr/include/x86_64-linux-gnu/qt5/QtGui -isystem /usr/include/x86_64-linux-gnu/qt5/QtWidgets -isystem /usr/include/x86_64-linux-gnu/qt5/QtConcurrent -isystem /usr/include/x86_64-linux-gnu/qt5/QtNetwork -g -O2 -ffile-prefix-map=/tmp/buildd/antimicro-3.1.4=. -fstack-protector-strong -Wformat -Werror=format-security -Wno-deprecated-declarations -Wno-deprecated-copy -Wdate-time -D_FORTIFY_SOURCE=2 -DANTIMICROX_PKG_VERSION=\"'3.1.4-1 (Debian sid/amd64)'\" -std=c++11 -fPIE -fPIC -o CMakeFiles/antimicrox.dir/src/eventhandlers/uinputeventhandler.cpp.o -c /tmp/buildd/antimicro-3.1.4/src/eventhandlers/uinputeventhandler.cpp
/tmp/buildd/antimicro-3.1.4/src/eventhandlers/uinputeventhandler.cpp: In member function ‘void UInputEventHandler::createUInputKeyboardDevice(int)’:
/tmp/buildd/antimicro-3.1.4/src/eventhandlers/uinputeventhandler.cpp:398:10: warning: ignoring return value of ‘ssize_t write(int, const void*, size_t)’ declared with attribute ‘warn_unused_result’ [-Wunused-result]
  398 |     write(filehandle, &uidev, sizeof(uidev));
      |     ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/tmp/buildd/antimicro-3.1.4/src/eventhandlers/uinputeventhandler.cpp: In member function ‘void UInputEventHandler::createUInputMouseDevice(int)’:
/tmp/buildd/antimicro-3.1.4/src/eventhandlers/uinputeventhandler.cpp:414:10: warning: ignoring return value of ‘ssize_t write(int, const void*, size_t)’ declared with attribute ‘warn_unused_result’ [-Wunused-result]
  414 |     write(filehandle, &uidev, sizeof(uidev));
      |     ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/tmp/buildd/antimicro-3.1.4/src/eventhandlers/uinputeventhandler.cpp: In member function ‘void UInputEventHandler::createUInputSpringMouseDevice(int)’:
/tmp/buildd/antimicro-3.1.4/src/eventhandlers/uinputeventhandler.cpp:438:10: warning: ignoring return value of ‘ssize_t write(int, const void*, size_t)’ declared with attribute ‘warn_unused_result’ [-Wunused-result]
  438 |     write(filehandle, &uidev, sizeof(uidev));
      |     ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/tmp/buildd/antimicro-3.1.4/src/eventhandlers/uinputeventhandler.cpp: In member function ‘void UInputEventHandler::write_uinput_event(int, int, int, int, bool)’:
/tmp/buildd/antimicro-3.1.4/src/eventhandlers/uinputeventhandler.cpp:459:10: warning: ignoring return value of ‘ssize_t write(int, const void*, size_t)’ declared with attribute ‘warn_unused_result’ [-Wunused-result]
  459 |     write(filehandle, &ev, sizeof(struct input_event));
      |     ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/tmp/buildd/antimicro-3.1.4/src/eventhandlers/uinputeventhandler.cpp:469:14: warning: ignoring return value of ‘ssize_t write(int, const void*, size_t)’ declared with attribute ‘warn_unused_result’ [-Wunused-result]
  469 |         write(filehandle, &ev2, sizeof(struct input_event));
      |         ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

While these are very minor, you might want to fix them (e.g. by signalling an error if the write fails, or (if errno == EINTR or only a partial block was written) even repeating it.

[mirabilos]

I’ve pushed a couple more issues (other than these I’ve mentioned here already) in #139, #140, #141, #142 and #143.

[mirabilos]

I’ve uploaded a package antimicro (= 3.1.4-1) containing AntiMicroX to Debian unstable; it should arrive there and be built for all architectures shortly.

As I do not have any usable joystick, PLEASE TEST THIS because I cannot do it beyond starting and testing some dialogues.

This WILL migrate to Debian testing, soon to be released as bullseye (11), in ten days, so please test before so, if anything should go wrong, we can block and fix it.

Once migrated, I can look at backporting the new version to Debian buster (10, current stable, soon to be oldstable). With this I can then also upload it to my PPA for *buntu bionic (18.04 LTS) and focal (20.04 LTS); other versions on request but nōn-LTS versions are best-effort and can only be used for a few months after release. The package should enter the normal *buntu distribution (hirsute, 21.04) a short time after it enters Debian and therefore will also be available there.

In case anyone wants to look at my packaging repository, feel free to.

I’d like to thank my employer ⮡ tarent @tarent for sponsoring work on packaging antimicro (a coworker explicitly requested it for a project) for Debian.

[pktiuk]

I have merged your changes, everything works properly with them applied.

[mirabilos]

Great, thanks.

Can you (or one of the devs anyway) please test the package as built in Debian, so we know that one’s also good?

[pktiuk]

I built it on fresh Debian 10 VM, but I was not able to run it.
For deb package built using cmake I get (this is the same problem I had in case of build for Raspberry Pi #121):

apt-get install ./antimicrox-3.1.4-x86_64.deb 
Reading package lists... Error!
E: Encountered a section with no Package: header

But in case of running bare binary it works properly.

[mirabilos]

No, you really must test the binaries as built by Debian; they most likely require Debian unstable, or at least Debian testing (11).

You also did not build the Debian package. The Debian package is called antimicro, not antimicrox.

wget https://deb.debian.org/debian/pool/main/a/antimicro/antimicro_3.1.4-1_amd64.deb
apt-get install "$PWD/antimicro_3.1.4-1_amd64.deb"

… is a quick-and-dirty way to do that.

[pktiuk]

Everything works well with this package.

[mirabilos]

OK, good to know, thanks! Then I’ll let it migrate to bullseye, buster-backports, *buntu PPA, …

[pktiuk]

Thanks for your help with this packaging :D

[gombosg]

Thanks for the incredible work you've done here @mirabilos ! :))

[mirabilos]

Backports and PPAs are filled with packages ☺ so I consider this finished from my side. A user reported in #89 that the backported package works well on his Raspberry, too.

Take-aways for you are still:

• the things I already opened separate issues for
• CHANGELOG.md is installed during build, twice even, but it isn’t needed even once as it’s embedded
• the headers are installed but they are now useless
• some compiler warnings in src/eventhandlers/uinputeventhandler.cpp

[pktiuk]

I think it can be closed now.

[mirabilos]

From reading 3.1.4...3.1.5 I don’t think the CHANGELOG.md is needlessly installed, twice, issue is fixed.

-- Installing: /tmp/buildd/antimicro-3.1.4/debian/antimicro/usr/share/doc/antimicrox/CHANGELOG.md
-- Installing: /tmp/buildd/antimicro-3.1.4/debian/antimicro/usr/share/antimicrox/CHANGELOG.md

Both these files are (almost certainly) not needed because the file is also embedded into the binary as Qt resource, as far as I can tell.

The -Wunused-result warnings in src/eventhandlers/uinputeventhandler.cpp are probably also still there?

[pktiuk]

From reading 3.1.4...3.1.5 I don’t think the CHANGELOG.md is needlessly installed, twice, issue is fixed.

-- Installing: /tmp/buildd/antimicro-3.1.4/debian/antimicro/usr/share/doc/antimicrox/CHANGELOG.md
-- Installing: /tmp/buildd/antimicro-3.1.4/debian/antimicro/usr/share/antimicrox/CHANGELOG.md

Both these files are (almost certainly) not needed because the file is also embedded into the binary as Qt resource, as far as I can tell.

Yes, one is embedded into the binary, but I am not sure and I don't want to delete it without checking (these two files seem to be source of problems earlier 8b695c7 8b695c7). - I will investigate it later as a part of #157

The -Wunused-result warnings in src/eventhandlers/uinputeventhandler.cpp are probably also still there?

I don't get any warning for uinputeventhandler (but I get others like -Wold-style-cast or -Wswitch).
Could you just paste this warning here?

[mirabilos]

Paweł Kotiuk dixit:

Yes, one is embedded into the binary, but I am not sure and I don't want to delete it without checking (these two files seem to be source

OK… build without it and test it the dialogue is still correctly shown?

- I will investigate it later as a part of #157

OK.

> The `-Wunused-result` warnings in > `src/eventhandlers/uinputeventhandler.cpp` are probably also still > there? I don't get any warning for uinputeventhandler (but I get others like `-Wold-style-cast` or `-Wswitch`). Could you just paste this warning here?

I did, see above. Thanks, //mirabilos --

emacs als auch vi zum Kotzen finde (joe rules) und pine für den einzig bedienbaren textmode-mailclient halte (und ich hab sie alle ausprobiert). ;)

Hallooooo, ich bin der Holger ("Hallo Holger!"), und ich bin ebenfalls ... pine-User, und das auch noch gewohnheitsmäßig ("Oooooooohhh"). [aus dasr]