yast/yast-installation#1077 (comment) demonstrates the correct appearance when using X11, whereas undermentioned is incorrect (default) colouration, expectedly inherited from the superuser:

PS /home/rokejulianlockhart> sudo lookandfeeltool --apply org.kde.breezedark.desktop
[sudo] password for root: 
qt.qpa.plugin: Could not load the Qt platform plugin "xcb" in "" even though it was found.
This application failed to start because no Qt platform plugin could be initialized. Reinstalling the application may fix this problem.

Available platform plugins are: eglfs, linuxfb, minimal, minimalegl, offscreen, vnc, wayland-egl, wayland, wayland-xcomposite-egl, wayland-xcomposite-glx, xcb.

NativeCommandExitException: Program "sudo" ended with non-zero exit code: 1.

Consider the following file:

{
  void f() {
    return nil;
  }
}

XML that ycpc produces when called by Y2R seems to contain wrong comment_before for <fun_def> — it contains two newlines, instead of just one:

<?xml version="1.0" encoding="UTF-8"?>
<ycp version="2.15.8.39280">
  <block kind="file">
    <symbols>
      <symbol global="1" category="filename" type="&lt;unspec&gt;" name="/tmp/y2r20130708-17363-etpazd"/>
      <symbol category="function" type="void ()" name="f"/>
    </symbols>
    <statements>
      <stmt><fun_def name="f" comment_before="&#xA;  &#xA;   ">
    <block kind="def" comment_before=" ">
      <statements>
        <stmt><return comment_before="&#xA;    " comment_after="&#xA;  "><yconst comment_before=" "><const type="void"/></yconst></return></stmt>
      </statements>
    </block>
</fun_def></stmt>
    </statements>
  </block>
</ycp>

It seems that part of comment_before is appended twice. If I replace function definition with structurally similar while statement, comment_before is correct:

{
  while (false) {
    y2milestone("m1");
  }
}

<?xml version="1.0" encoding="UTF-8"?>
<ycp version="2.15.8.39280">
  <block kind="file">
    <symbols>
      <symbol global="1" category="filename" type="&lt;unspec&gt;" name="/tmp/y2r20130708-18310-90tn7r"/>
    </symbols>
    <statements>
      <stmt><while comment_before="&#xA;  " comment_after="&#xA;">
  <cond><yconst comment_before=" "><const type="bool" value="false"/></yconst></cond>
  <do><block kind="stmt" comment_before=" ">
  <statements>
    <stmt><builtin name="y2milestone" comment_before="&#xA;    " comment_after="&#xA;  " args="1"><element><yconst><const type="string" value="m1"/></yconst></element></builtin></stmt>
  </statements>
</block>
  </do>
</while></stmt>
    </statements>
  </block>
</ycp>

When executed in comment-enabled mode, some unusual (but legal) whitespace characters are emitted unescaped into XML comment_before and comment_after attributes. One such example is form-feed (0x0C), which caused trouble in Y2R because Nokogiri didn't parse the XML which contained it correctly.

If fixed all form feed cahracter occurences in the following pull requests:

• yast/yast-network#98
• yast/yast-runlevel#10

It would still be good to fix the XML serializer to skip all such characters (as they apparently can't be included even by reference like ). See http://www.w3.org/TR/xml/#charsets for the list of allowed ones.

YCP code

float x = 0.9;

is wrongly converted to

<assign name="x"><const type="float" value="0."/></assign>

The decimal part is missing in the output.

While working on reproducible builds for openSUSE, I found that yast2-core fails to build in 2038.

# of expected passes           522
# of unexpected failures       1

To reproduce, use

osc build --vm-type=kvm --clean --noservice --build-opt=--vm-custom-opt="-rtc base=2038-01-25T00:00:00" --alternative-project home:bmwiedemann:reproducible openSUSE_Tumbleweed

It writes 0,5 instead of 0.5:

$ cat half.ycp 
{
  float half = 0.5;
}
$ LANG=cs_CZ.UTF-8 ycpc -c -x -o /dev/stdout half.ycp |& grep assign
  <stmt><assign name="half"><const type="float" value="0,5"/></assign></stmt>

One consequence is a failing test yast2/library/types/testsuite/tests/String.rb

Given the following YCP code:

{list l=[/*b1*/1/*a1*/,/*b2*/2/*a2*/,/*b3*/3/*a3*/];}

I get the following XML from ycpc in comment-enabled mode:

<?xml version="1.0" encoding="UTF-8"?>
<ycp version="2.15.8.39280">
  <block kind="file">
    <symbols>
      <symbol global="1" category="filename" type="&lt;unspec&gt;" name="/tmp/y2r20130711-8472-1qg7ee"/>
      <symbol category="variable" type="list" name="l"/>
    </symbols>
    <statements>
      <stmt><assign name="l" comment_before=" "><list size="3" comment_before="/*a3*/"><element><yconst comment_before="/*b1*/"><const type="int" value="1"/></yconst></element><element><yconst comment_before="/*b2*/"><const type="int" value="2"/></yconst></element><element><yconst comment_before="/*b3*/"><const type="int" value="3"/></yconst></element></list></assign></stmt>
    </statements>
  </block>
</ycp>

Comments /*a1*/ and /*a2*/ are missing.

To facilitate implementing YaST modules in various languages, yast2-core loads the optional language binding libraries, /usr/lib*/YaST2/plugin/libpy2lang_*.so, packaged in yast2-*-bindings.rpm.

But these are loaded even when we have compiled an incompatible core (eg. lexer_comments, because of adding comment members to YCode) and want to run that from the working copy. This is the cause of @dmajda's crash. We've worked around it by using ./configure --prefix=/nosuchpath in his WC. (Another workaround is to uninstall the bindings)

Also, IIUC, this was the cause of the crashes in Jenkins last week. @jreidinger can you confirm that and comment on my proposed solution?:

I think we should have a switch to enable/disable the language plugin loading. It should stay enabled by default for compatibility. I propose an environment variable, Y2DISABLELANGUAGEPLUGINS=..., to be checked in Y2LanguageLoader::Y2LanguageLoader ().

Fails due to the following header changes:
https://gcc.gnu.org/gcc-13/porting_to.html#header-dep-changes

$ osc co openSUSE:Factory/yast2-core
$ cd $_
$ osc build --alternative-project devel:gcc:next
...
[   26s] y2changes.cc: In function 'std::string y2_changesfmt_prefix(logcategory_t)':
[   26s] y2changes.cc:135:24: error: 'time' was not declared in this scope; did you mean 'utimes'?
[   26s]   135 |     time_t timestamp = time (NULL);
[   26s]       |                        ^~~~
[   26s]       |                        utimes
[   26s] y2changes.cc:136:29: error: 'localtime' was not declared in this scope
[   26s]   136 |     struct tm *brokentime = localtime (&timestamp);
[   26s]       |                             ^~~~~~~~~
[   26s] y2changes.cc:138:5: error: 'strftime' was not declared in this scope; did you mean 'st_ctime'?
[   26s]   138 |     strftime (date, sizeof (date), Y2CHANGES_DATE, brokentime);
[   26s]       |     ^~~~~~~~
[   26s]       |     st_ctime
[   26s] make[4]: *** [Makefile:569: y2changes.lo] Error 1

Consider the following YCP code:

{
  textdomain "foo";

  string s = /* before */ _("bar");
}

When I run ycpc -cx on it with Y2PARSECOMMENTS=1, I get the following XML:

<?xml version="1.0" encoding="UTF-8"?>
<ycp version="2.15.8.39280">
  <block kind="file">
    <symbols>
      <symbol global="1" category="filename" type="&lt;unspec&gt;" name="../testcases/units.ycp"/>
      <symbol category="variable" type="string" name="s"/>
    </symbols>
    <statements>
      <stmt><textdomain name="foo" comment_before=" &#xA;  "/></stmt>
      <stmt><assign name="s" comment_before=" &#xA;&#xA;   " comment_after="&#xA;"><locale domain="foo" text="bar"/></assign></stmt>
    </statements>
  </block>
</ycp>

The XML does not contain the /* before */ comment.

The minimal testcase comes form String.ycp.

With yast2-core from YaST:Head:ruby, I found this while typing this into @kobliha's y2r-editor:

{
  // unterminated string before EOF
  any hello = "World!

It hangs the editor, and ycpc -c shows an endless stream of

snippets/string-loop.ycp:4  [Parser] unterminated string constant

Only kill -9 helped.

Working around it by disabling the test for now (#31).

Should enable verbose test runs and ignore failures to find out HOW it fails.

Hello,
Is yast4debian ready for production site? https://github.com/yast4debian
Regards, Heather

// Write a value into file
SCR::Write(".network.value.\"eth0\".IPADDR", "1")
SCR::Write(".network", nil)

// delete the file
SCR::Write(".network.section.\"eth0\"", nil)
SCR::Write(".network", nil)

In this case, /etc/sysconfig/network/ifcfg-eth0 should be written (it is) and removed subsequently. But when the script is done, the file is still present.