Mayavi2 is just TRULY awesome in terms of capabilities. However, this is a general issue... and I have huge trouble every time I want to install mayavi
in ubuntu
inside virtualenv
or on the core system. It is just not a clean 'pip' install even after all vtk
etc. dependencies are installed in core. Dependencies and options are just too many and too varied. Many packages it requires are not identified automatically. Countless people have seen these problem - and every-time I have to go looking for solutions online. May be I am doing something wrong... but this has always been the case. May be mayavi
should think about just limiting the backend options to just a few but have it all really cleanly install-able and easily usable without problems. Will it ever be possible to ever install mayavi
with just a simple command pip install mayavi
and work completely ready out-of-box without any errors? I will be glad to help the owners in development. I understand vtk
etc. dependencies, however looks like it is now available through pip
.
After single pip
install, all options could be simply controllable via environmental variable... For example, Use of mesa
or xvfb
. One install should do everything, possibly; and everything else should be controllable via a simple .mayavi
startup file or environment variables. There may be just multiple copies of the libraries which are differently compiled under their respective trees... or an option to do so by the user.
mayavi
does super heavy-duty lifting, but the true glory will be even greater in super easy install. Not the only option... but pip
may be the best way to do this well and completely.
For example, a fresh install and I see this error multiple times:
QVariantMap DBusMenuExporterDBus::getProperties(int, const QStringList&) const: Condition failed:
action
@Kekushke -- I am working on this and I was the one who pushed the VTK wheels to PyPI. Some of your errors are because you are using the last released version of Mayavi. pip install will not work as I need time to make a release. Unfortunately, I have a problem this weekend which has meant staying away from my computer and over the rest of the week, end semester work. I was hoping to release mayavi this weekend or the next but I am afraid it is going to have to wait until I feel better and I am able to get some time. I am hoping by next weekend to have a release.
@prabhuramachandran Thank you so much for your comments. Hope you feel better soon. No hurry at all. I just wanted to point of users' perspective and experience; hoping to help. This is an awesome awesome tool - and as the user experience gets better this is going to be here for another 10-, 25-years and beyond!
It is amazing that you can do so much with mayavi
and there are just numerous options... and I think latter has added too much complexity, but may be there are ways to simplify things so that it is easy to choose those options with ultra-simple installation process . Everything in mayavi
works, but user has to just fight it out... and that could go away. I will be glad to help test/write docs/tutorials and simplify things if you wish.
Thanks again!
@Kekushke -- please don't call me Prof., I understand your pain and am trying to make things a bit easier to install. Thank you for your kind words and encouragement. I also have other tools I have been maintaining and perhaps I have bitten off more than I can chew! Any help is welcome. I think if I just push a new Mayavi release, Mayavi will be fully pip installable which will help loads. Hopefully by the coming weekend.
@prabhuramachandran No problem! Looking at the history of mayavi
, I figured you are the creator! Really awesome! Thanks again! :-)
@prabhuramachandran Mayavi is awesome! I am trying to animate some quantum dynamics simulations and this is the only tool I know which will let me do what I want, so thank you. But I have to ask, why is the dependency tree such a nightmare? I've been trying for a day and half to get it working on my new MacBook Pro and the sequence of errors is unbelievable. It probably doesn't help that this issue has existed for several years and so I keep finding out of date advice, but honestly, this is not sustainable.
I understand but the following steps should suffice on the three major platforms (Python 3.x), with all the different Python distributions:
pip install vtk traits traitsui pyface envisage apptools pyqt5
git clone https://github.com/enthought/mayavi.git
cd mayavi
python setup.py install
In a few days I hope to push a new release and that should make it possible to simply do a pip install mayavi
.
Thank you, that actually worked! You should put these clear instructions in the README
I think this issue is now resolved so closing. Thanks for your patience.
Most helpful comment
I understand but the following steps should suffice on the three major platforms (Python 3.x), with all the different Python distributions:
In a few days I hope to push a new release and that should make it possible to simply do a
pip install mayavi
.