New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
drop eos for boost 106900 #178
Conversation
since boost 1.69 drops floating point traits class use boost binary/polimorphic ioarchive instead
Consider, if |
Lux uses IEEE745 full feature set and, for instance, compiling with "fast-math" option (i.e. relaxing some of the 745 constrains) introduces many subtle bugs. This may be a problem if switching to binary archives too. Binary archives are also not portable across different CPU architectures (i.e. little endian Vs big endian). So just switching to normal binary archives, it is not a viable solution. Latest Boost documentation mentions a "portable binary archives" example: https://www.boost.org/doc/libs/1_69_0/libs/serialization/doc/todo.html#portablebinaryarchives The examples are available under boost_1_69_0/libs/serialization/example and they may be like the correct way to fix this problem. Looking at the sources, they seems to only address the little Vs. big endian problem but it would be a start point. |
Has this stagnated? What are the plans for this now? The issue this is addressing, is preventing to build on my system (and I'm not going to downgrade my boost for LuxCore). I'd rather have this (possibly with a |
This is rejected as explained in my previous post.
You shouldn't not build LuxCore with a different Boost version. It is not supported and can (and have in the past) easily introduced many subtle and not subtle bugs. The only procedure to Build LuxCore we support is described in https://github.com/LuxCoreRender/LinuxCompile and https://github.com/LuxCoreRender/WindowsCompile They are both static build methods and don't require to downgrade/upgrade anything on your system. |
Noticed you've dropped EOS. There's an EOS compatibility patch for boots v1.72 in CERN-CMS project perhaps you could use it here if some problem pops up. |
Yes, we are in the process of upgrading Boost library to v1.72. I'm simply giving up in supporting binary files interoperability between CPUs with different FP32 formats. I prefer to not add another mostly not maintained EOS-like dependency. |
since boost 1.69 drops floating point traits class
use boost binary/polymorphic archive instead.
solves #169