ratt-ru / meqtrees

A library for implementing radio astronomical Measurement Equations
http://meqtrees.net
10 stars 2 forks source link

Hippo draw / Histogram of cache result #99

Closed gijzelaerr closed 10 years ago

gijzelaerr commented 10 years ago
at 2005-03-10 12:31:23 Rob Assendorp reported:

Hippo draw / Histogram of cache result

gijzelaerr commented 10 years ago

Original comment thread migrated from bugzilla

at 2005-03-10 12:31:23 Rob Assendorp replied:

hippo draw is no longer available in the browser. Can we get it back? An alternative would be to have a histogram option in the current result plotter.

at 2005-03-10 16:22:01 Oleg Smirnov replied:

OK, you need to fix the hippo plugin and I need to get hippo running on lofar9/10. Please reassign this to Hardware/OS and me when you've got the plugin fixed.

at 2005-03-10 21:07:33 Tony Willis replied:

Hi Oleg

I have checked in Plugins/hippo_array_plotter.py and updated meqbrowser.py.

However, your new Grid System may have killed Hippo in any case! I now get a segmentation fault on my laptop when I create the Tuple data structure that holds the incoming data. I'm curious if the same seg fault occurs on lofar9 when you install Hippo there. If not I guess the problem lies with the various c++ compilers I've used. If things croak on lofar9, I guess we should set a priority for implementation of histograms with Qwt.

at 2005-03-11 19:33:17 Tony Willis replied:

Hippo sucks! (At least on Tony's laptop). For reasons known only to itself, Hippo 1.12.2 compiled with gcc 3.2.2 now seg faults when called from the python interface to Oleg's new browser as it tries to instantiate an ostrstream object! Attempts to compile Hippo with gcc3.4 fail because the configure script thinks there is a problem with the Qt library interface!

Bleagh!

at 2005-03-11 20:59:43 Oleg Smirnov replied:

The segfault is to be expected, Hippo uses libstdc++ and so do some meqbrowser components, so to link them together you must have them both compiled against the same generation of libstdc++, which unfortunately changed between gcc-3.3 and 3.4.

Why it won't compile with gcc-3.4, I dunno... my sympathies... for extra self-inflicted torture points, have you tried rebuilding Qt and sip with gcc-3.4?

at 2005-03-12 01:58:31 Tony Willis replied:

Yes - I compiled Qt with gcc 3.4 hoping the Hippo configure error would go away, but it doesn't, at least on my system. So maybe I still have some incompatibilities somewhere - I'll be interested if it configures on lofar 9 / 10.

I did install numarray-1.2.3 on my laptop, and after re-compiling PyQwt the meqbrowser continued to work! Since this version of numarray has a built-in histogram function, it may not be much work to get a histogram and plot it with Qwt (the qwt developer has promised to send me some c++ code which I can translate into python.

at 2005-03-15 21:45:39 Tony Willis replied:

In order to get rid of our dependence on Hippo I have now created a simple Histogram plot plugin based on Qwt. The initial version is quite primitive - no hardcopy, zooming, etc, yet and the number of bins is set at 10, but at least something will be shown. If you right-click on an array one of the options now shown is 'Histogram Plot'

Cheers

Tony

at 2005-03-20 00:19:08 Oleg Smirnov replied:

Given all the recent problems building Hippo, it hardly seems worth the effort maintaining it (not to mention the ton of other system dependencies -- bjam, boost, etc. -- that it drags onboard). Especially now than Tony's created a Qwt-based histogram view. Therefore I'm closing this bug; if someone would like to see Hippo-based viewers again, they can always reopen it.