Re: [platform-testers] texinfo pretest

[Gavin Smith]

On 17 May 2015 at 23:22, Assaf Gordon <address@hidden> wrote:
> Hello Gavin,
>
> Few more tests with the SVN r6275.
>
> -assaf
>
>
> ---- OpenBSD 5.6:
> the first few tests pass, and 't/dir-dangling-entry.sh' is skipped,
> until 't/goal-column.sh' which is stuck.
>
> Based on your suggestion to open a new terminal and print the log:
>
>     $ ls -l goal-column.sh*
>     -rwxr-xr-x  1 miles  wheel  977 May 16 16:53 goal-column.sh
>     prw-r--r--  1 miles  wheel    0 May 17 17:38 goal-column.sh.finished
>     -rw-r--r--  1 miles  wheel  507 May 17 17:41 goal-column.sh.log
>     prw-r--r--  1 miles  wheel    0 May 17 17:38 goal-column.sh.pipein
>     prw-r--r--  1 miles  wheel    0 May 17 17:38 goal-column.sh.pipeout
>     $ cat goal-column.sh.log
>     pseudotty: getting pty master fd
>     pseudotty: unlocking slave device
>     pseudotty: getting file name of slave device...
>     pseudotty: /dev/ttyp1
>     pseudotty: entering main loop
>     pseudotty: read error on master fd
>     Fetching PID of ginfo process under test...
>     ...unknown
>     ./t/goal-column.sh[23]: cannot create t/goal-column.sh.pipeout:
> Interrupted system call
>     ./t/goal-column.sh[21]: cannot create /dev/ttyp1: Interrupted system
> call
>     ./t/goal-column.sh[24]: cannot create t/goal-column.sh.pipeout:
> Interrupted system call

I think this is likely caused by the same problem as the
dir-dangling-entry.sh test hanging. The "read error on master fd"
message suggests that "ginfo" has exited, leading pseudotty to
disappear taking the *.pipeout file with it.

Thanks to the OpenCSW buildfarm I think I've fixed this.

> --- OpenSUSE 13.2:
>
> Still two test fails:
>    FAIL: t/c-u-m-x-scroll-forward.sh
>    FAIL: t/index-apropos.sh
>
> But perhaps this is relevant: there are also five expected XFAILS:
>    XFAIL: t/quoted-label-as-target.sh
>    XFAIL: t/quoted-target.sh
>    XFAIL: t/quoted-label-and-target.sh
>    XFAIL: t/goto-quoted.sh
>    XFAIL: t/next-quoted.sh
>
> and in the log, it seems both the FAIL and XFAIL print the same errors. so
> perhaps that should be expected?
> log attached.

The logs of the XFAIL tests look right to me. I think the most likely
explanation of the FAIL: t/c-u-m-x-scroll-forward.sh is that a byte is
going missing somewhere. I don't know about index-apropos.sh. All I
can suggest is trying it again and looking at the logs?