Closed IanHeywood closed 8 years ago
Could you do a dpkg -la | grep blitz
for me please? This is the same problem as https://github.com/ska-sa/meqtrees-timba/issues/18.
Should look like this BTW:
$ dpkg -la | grep blitz
ii libblitz-doc 1:0.10-1ubuntu1 all C++ template class library for scientific computing
ii libblitz0-dev 1:0.10-1ubuntu1 amd64 C++ template class library for scientific computing
ii libblitz0ldbl 1:0.10-1ubuntu1 amd64 C++ template class library for scientific computing
...but there was a 0.11 package floating around the repos that's actually an older blitz which would cause the above error.
ianh@nerv:~/meqtrees/meqtrees-timba/build/release$ dpkg -la | grep blitz
ii libblitz0-dev 1:0.11-1trusty amd64 C++ template class library for scientific computing
ii libblitz0ldbl 1:0.11-1trusty amd64 C++ template class library for scientific computing
Yep that's the culprit. Purge it, then apt-get install libblitz0-dev
. You need the bog-standard Ubuntu trusty 1:0.10 version.
But before you do, can you try $ apt-show-versions -a libblitz0-dev
? I'd like to figure out where this 1:0.11 package got installed from. We need to find it and repress it...
ianh@nerv:~/meqtrees/meqtrees-timba/build/release$ apt-show-versions -a libblitz0-dev
libblitz0-dev:amd64 1:0.11-1trusty install ok installed
libblitz0-dev:amd64 1:0.10-1ubuntu1 trusty au.archive.ubuntu.com
libblitz0-dev:amd64 1:0.11-1trusty trusty ppa.launchpad.net
libblitz0-dev:amd64/trusty 1:0.11-1trusty uptodate
So I can apt-get remove plus the package names that dpkg returned?
I think this is an old system and got it from ska-sa ppa when it was still available there.
I removed all packages from ska-sa ppa just to be sure. Op 2 sep. 2015 3:00 PM schreef "Oleg Smirnov" notifications@github.com:
Yep that's the culprit. Purge it, then apt-get install libblitz0-dev. You need the bog-standard Ubuntu trusty 1:0.10 version.
But before you do, can you try $ apt-show-versions -a libblitz0-dev? I'd like to figure out where this 1:0.11 package got installed from. We need to find it and repress it...
— Reply to this email directly or view it on GitHub https://github.com/ska-sa/meqtrees/issues/868#issuecomment-137091049.
You should apt-get remove the blitz packages, then apt-get update just to be sure, then apt-get install libblitz0-dev. Check the output to make sure that 1:0.10 is being installed not 1:0.11. If it still gets 1:0.11 somewhere, then @gijzelaerr needs to do some more scouring of the repos...
OK it's installed 1:0.10 across the board from your method. Rebuilding now...
Thanks.
Let us hope 1:0.11 has been properly Stalinized from all repositories.
Can confirm that this works. I've also edited the suggested .bashrc items on the building from source page which didn't reflect current paths.
Excellent, thanks!
On Wed, Sep 2, 2015 at 4:40 PM, IanHeywood notifications@github.com wrote:
Can confirm that this works. I've also edited the suggested .bashrc items on the building from source page which didn't reflect current paths.
— Reply to this email directly or view it on GitHub https://github.com/ska-sa/meqtrees/issues/868#issuecomment-137107131.
Likewise! Cheers.
I recently went through the exercise of building casacore. casarest, lwimager, and wsclean on the Cambridge supercomputer. Only took about 4 days of headbanging to get things right. The final joke of course was that using a supercomputer doesn't help that much because of remote disk I/O issues with CASA measurement sets. A wsclean that took about 1.5 hours on my laptop timed out after 5 hours wallclock time on the Cambridge computer. Sigh.
Apologies for the slightly off topic message. I’ve been compiling the same stack recently and had similar issues. But seriously this email MADE MY DAY!!! ROTFL!!!
On 2 Sep 2015, at 20:04, Tony Willis notifications@github.com wrote:
I recently went through the exercise of building casacore. casarest, lwimager, and wsclean on the Cambridge supercomputer. Only took about 4 days of headbanging to get things right. The final joke of course was that using a supercomputer doesn't help that much because of remote disk I/O issues with CASA measurement sets. A wsclean that took about 1.5 hours on my laptop timed out after 5 hours wallclock time on the Cambridge computer. Sigh.
— Reply to this email directly or view it on GitHub.
Up to the 'Building the software' section on this page. Brace yourselves...