[prev in list] [next in list] [prev in thread] [next in thread] 

List:       macports-users
Subject:    Re: PortIndex failing for a few ports due to hdf5 not active
From:       David Strubbe <dstrubbe () macports ! org>
Date:       2015-11-20 4:15:13
Message-ID: CAENVOuWvY6YvarbNdai2+xjPSKR1Amd7DbBbiSXw-StsPMLxtQ () mail ! gmail ! com
[Download RAW message or body]

[Attachment #2 (multipart/alternative)]


Yes this was due to a mistake in r142662, now fixed.

However, it is also due to the fact that cdo and ncarg seem to need hdf5
but do not declare it as a dependency.

Additionally, cdo, ncarg, and hdfeos5 all use the mpi portgroup strangely
(e.g. below), in trying to check how hdf5 was installed when parsing the
Portfile. Is one supposed to be able to do this sort of thing?

set v [mpi_active_variant_name bin:h5pcc:hdf5]
if {${v} != "" } {
    depends_build-append    port:yaxt
    configure.cc            ${prefix}/bin/h5pcc
}

The mpi portgroup has this comment, in mpi.enforce_variant, when the mpi
variants do not match:
                # instead of trying to append to a default variant (which
fails
                # silently for some cases), we should be explicit and tell
the
                # user about the error
                ui_error "Need to select variant +$otmpi"
which suggests to me that what is being done in these Portfiles may be
problematic.

David


On Thu, Nov 19, 2015 at 1:39 PM, Ryan Schmidt <ryandesign@macports.org>
wrote:

>
> On Nov 19, 2015, at 11:34 AM, Adam Mercer wrote:
>
> > When I recently updated a 10.11 machine I noticed the following during
> > a port sync:
> >
> > Creating port index in
> > /opt/local/var/macports/sources/
> rsync.macports.org/release/tarballs/ports
> > Failed to parse file science/cdo/Portfile: Registry error: hdf5 not
> > registered as installed & active.
> > Failed to parse file science/hdfeos5/Portfile: Registry error: hdf5
> > not registered as installed & active.
> > Failed to parse file science/ncarg/Portfile: Registry error: hdf5 not
> > registered as installed & active.
> >
> > Surely generating the PortIndex shouldn't require that dependent ports
> > are installed?
>
> Sounds like a bug in those ports -- or a portgroup they include: the mpi
> 1.0 portgroup, which was recently updated. Perhaps that update caused this?
>
>
>

[Attachment #5 (text/html)]

<div dir="ltr">Yes this was due to a mistake in r142662, now \
fixed.<div><br></div><div>However, it is also due to the fact that cdo and ncarg seem \
to need hdf5 but do not declare it as a \
dependency.</div><div><br></div><div>Additionally, cdo, ncarg, and hdfeos5 all use \
the mpi portgroup strangely (e.g. below), in trying to check how hdf5 was installed \
when parsing the Portfile. Is one supposed to be able to do this sort of \
thing?</div><div><br></div><div><div>set v [mpi_active_variant_name \
bin:h5pcc:hdf5]</div><div>if {${v} != &quot;&quot; } {</div><div>      \
depends_build-append      port:yaxt</div><div>      configure.cc                  \
${prefix}/bin/h5pcc</div><div>}</div></div><div><br></div><div>The mpi portgroup has \
this comment, in mpi.enforce_variant, when the mpi variants do not \
match:</div><div><div>                        # instead of trying to append to a \
default variant (which fails</div><div>                        # silently for some \
cases), we should be explicit and tell the</div><div>                        # user \
about the error</div><div>                        ui_error &quot;Need to select \
variant +$otmpi&quot;</div></div><div>which suggests to me that what is being done in \
these Portfiles may be \
problematic.</div><div><br></div><div>David</div><div><br></div></div><div \
class="gmail_extra"><br><div class="gmail_quote">On Thu, Nov 19, 2015 at 1:39 PM, \
Ryan Schmidt <span dir="ltr">&lt;<a href="mailto:ryandesign@macports.org" \
target="_blank">ryandesign@macports.org</a>&gt;</span> wrote:<br><blockquote \
class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc \
solid;padding-left:1ex"><span class=""><br> On Nov 19, 2015, at 11:34 AM, Adam Mercer \
wrote:<br> <br>
&gt; When I recently updated a 10.11 machine I noticed the following during<br>
&gt; a port sync:<br>
&gt;<br>
&gt; Creating port index in<br>
&gt; /opt/local/var/macports/sources/<a \
href="http://rsync.macports.org/release/tarballs/ports" rel="noreferrer" \
target="_blank">rsync.macports.org/release/tarballs/ports</a><br> &gt; Failed to \
parse file science/cdo/Portfile: Registry error: hdf5 not<br> &gt; registered as \
installed &amp; active.<br> &gt; Failed to parse file science/hdfeos5/Portfile: \
Registry error: hdf5<br> &gt; not registered as installed &amp; active.<br>
&gt; Failed to parse file science/ncarg/Portfile: Registry error: hdf5 not<br>
&gt; registered as installed &amp; active.<br>
&gt;<br>
&gt; Surely generating the PortIndex shouldn&#39;t require that dependent ports<br>
&gt; are installed?<br>
<br>
</span>Sounds like a bug in those ports -- or a portgroup they include: the mpi 1.0 \
portgroup, which was recently updated. Perhaps that update caused this?<br> <br>
<br>
</blockquote></div><br></div>



_______________________________________________
macports-users mailing list
macports-users@lists.macosforge.org
https://lists.macosforge.org/mailman/listinfo/macports-users


[prev in list] [next in list] [prev in thread] [next in thread]