Created attachment 365990 [details]
On my system 'emerge -c' fails due to the following error:
* Dependencies could not be completely resolved due to
* the following required packages not being installed:
* sci-libs/hdf5:0/1.8.11= pulled in by:
Manually downgrading hdf5 to version 1.8.11 temporarily resolves this, but upon 'emerge -uD @world' it gets upgraded to version 1.8.12 again.
I'll try to give the relevant info, but tell me if I miss something. Output of 'emerge -vcp hdf5':
Calculating dependencies... done!
sci-libs/hdf5-1.8.12 pulled in by:
dev-python/pytables-3.0.0 requires sci-libs/hdf5:=
media-libs/vigra-1.9.0 requires sci-libs/hdf5
sci-libs/flann-1.8.4 requires sci-libs/hdf5
pytables is ultimately pulled in by sci-mathematics/sage on my system (although sage is not in main tree, the ebuilds for hdf5 and pytables are from the main tree).
What's the output of "emerge -cp" when hdf-1.8.11 is installed?
Please attach the debug.log created by
"emerge -uD @world --pretend --debug &>debug.log"
when hdf-1.8.11 is installed.
(In reply to Sebastian Luther (few) from comment #1)
> What's the output of "emerge -cp" when hdf-1.8.11 is installed?
It says everything is fine and nothing needs to be removed.
> Please attach the debug.log created by
> "emerge -uD @world --pretend --debug &>debug.log"
> when hdf-1.8.11 is installed.
In the works.
Created attachment 366000 [details]
requested output of emerge --debug (xzipped)
HTH, it's rather large ...
Please attach the depclean_debug.log created by
"emerge -cpv --debug &> depclean_debug.log".
According to your debug.log dev-python/pytables isn't a deep dependency of @world. This explains why its dependency on hdf5 is ignored with -uD. I would expect it to be removed by "emerge -c". Lets find out why it isn't.
Created attachment 366052 [details]
requested output of emerge --depclean --debug (xzipped)
I manually pieced together the following dependency chain:
and I took a close look at my world file and sage is listed normally like every other atom.
This seems to be a duplicate of bug 493374.
The problem is that the dependency between rpy and pandas is only build time.
This leads emerge to ignore it during -uD world, but consider it during -c.
As a workaround use either --with-bdeps=y for -uD or --with-bdeps=n for -c.
*** This bug has been marked as a duplicate of bug 493374 ***
I changed my mind. The issue is different and may be fixed in a different way.
17 Feb 2015; Justin Lecher <firstname.lastname@example.org> -pytables-3.0.0.ebuild,
(In reply to Sebastian Luther (few) from comment #6)
> This seems to be a duplicate of bug 493374.
> The problem is that the dependency between rpy and pandas is only build time.
> This leads emerge to ignore it during -uD world, but consider it during -c.
> As a workaround use either --with-bdeps=y for -uD or --with-bdeps=n for -c.
Unfortunately, there's no way do avoid inconsistencies like this when you have an option like --with-bdeps=n, so I recommend to put --with-bdeps=y in EMERGE_DEFAULT_OPTS if it bothers you.