[pkg-firebird-general] Bug#874517: firebird3.0 FTCBFS: lots of issues in configure.ac

[Helmut Grohne]

Hi Damyan,

On Thu, Sep 07, 2017 at 01:20:03PM +0000, Damyan Ivanov wrote:
> Thanks. That failed for me when building natively with
> 
>   unsupported platform
> 
> I applied the following additional patch in order to get it to build 
> natively:
> 
> --- a/configure.ac
> +++ b/configure.ac
> @@ -85,7 +85,9 @@ STD_EDITLINE=false
>  dnl Test for special ar options?
>  AR_OPT_CHECK=false
>  
> +if test -z "$host" ; then host="$build"; fi
> +
>  host_quadruplet=$(echo "$host" | sed 's/^\(@<:@^-@:>@*\)-\(@<:@^-@:>@*\)-\(@<:@^-@:>@*\)$/\1-pc-\2-\3/')
>  echo "considering >$build< >$host< >$host_quadruplet<"
>  case "$host_quadruplet" in
> @@ -484,7 +488,7 @@ dnl CPU_TYPE=ppc64
>      ;;
>  
>    *)
> -    AC_MSG_ERROR(unsupported platform ${build})
> +    AC_MSG_ERROR(unsupported platform ${host_quadruplet})
>      ;;
>  esac

I'm sorry. I should have checked native building. Thanks for fixing up
my patch.

> lockfile usage was introduced in a debian patch (since firebird 2.5) 
> that enables building in parallel with -j. It was needed because parts 
> of the upstream build system use the just compiled firebird database 
> engine, and that usage needed to be serialized.

I understood all that at the time of filing the report, because it was
well-documented in said Debian patch. Thanks!

> In firebird 3.0, however, it seems the engine copes with being used in 
> parallel, so I dropped the patch, as well as the procmail 
> build-dependency.

Cool. I was looking to the correct Multi-Arch annotation of procmail,
but it seems I can skip that now. :)

> However, I think that this will not solve the cross-building problem, 
> since the upstream build still relies on executing just built 
> binaries, e.g. gpre -- a source preprocessor, and the engine itself.

I suggest that you really close this bug with the upload.  Thus I will
notice when your changes reach unstable. If I come up with a patch for
the other issues, I can still make new report. I don't believe that
FTCBFS bugs without patches serve any purpose beyond cluttering your
view.

Alternatively, I'd like to receive a poke when this hits unstable, such
that I can schedule a retest manually.

Helmut