peterspackman / crystalexplorer

Crystal structure analysis with Hirshfeld surfaces, intermolecular interaction energies and more
https://crystalexplorer.net
GNU Lesser General Public License v3.0
4 stars 1 forks source link

I had some problems compiling the new tonto-free version #20

Open MilitaoLucas opened 3 weeks ago

MilitaoLucas commented 3 weeks ago

Hello, good morning. Trying the new version (I know this is very early, but this issue is relevant) I found some problems in unique_ptr in std::unique_ptr. I am not very familiar with C++ (working on getting better) but I found in StackOverflow that including the <memory> header seems to fix this problem. I was able to compile and run the new version this way. Other problem is that you didn't include OCC_PATH in the header file of settings.cpp. I was able to compile by providing:

const QString OCC_PATH = OCC_GROUP + "";

But I don't know what the default should actually be, so I didn't fell comfortable creating a PR.

peterspackman commented 3 weeks ago

Thanks for pointing this out - if you could mention which header file(s) were missing an include from the <memory> header that would be helpful!

The rework branch at the moment is 100% not ready for others to use (or probably even compile) and has many unresolved issues. I'm aiming to get an early release version ready within the next month.

MilitaoLucas commented 3 weeks ago

Oh, I totally forgot that. My bad. The new version had some problems, like looking for absolute paths, but I was able to work trough that. The results are looking similar to the older version, and I am liking it very much. Thank you for the awesome work! kdtree.h seems to be missing the <memory> header. Also, the OCC path is missing from settings.h. I decided to compile and use the new version because I had some problems in doing anything using the HOZPIX refcode in both Linux (I used the CentOS build) and on a Windows VM (CrystalExplorer just crashed).

MilitaoLucas commented 3 weeks ago

In the other topic, I was able to use HOZPIX in the main version fine. But I liked the new version better, and after working through the issues it seemed to work. The website version had problems working with this specific cif in both the Windows VM and Linux.

peterspackman commented 3 weeks ago

Glad to hear things are working well enough to be useful - it's still very much a work in progress but there should be a lot of improvements moving forward and the code being overhauled should make extensions/additions much easier than it was before.

I still haven't sorted out the OCC_PATH variabls etc. which need to be basically determined by the CMake configuration (i.e. install location). That is relatively easy to fix and I'll get to it when I'm fixing up all the packaging.

Just bear with me while I get things finished up to where they're good enough for people to start alpha testing!

MilitaoLucas commented 3 weeks ago

Sure, I was just trying a little too much to make it work, so I could still include it in my undergrad thesis. Sorry for creating an issue for pre-alpha code.