On 10/22/21 9:23 PM, Étienne Mollier wrote:
Hi Nilesh, Nilesh Patra, on 2021-10-22:looks like C++17 has removed the provision to specify custom exceptions, and there are a lot of errors like the one above. In this case, does it makes sense to force C++14 standard (with chnages in d/rules or with patches) in such cases and ask upstream to clean these up? Or do I miss a relatively easy solutions?It depends. If I read correctly the porting guide to Gcc 11 [1] and if the number of occurrence of the issue is reasonable, then you can just patch the throw(BadOptionSetting) occurrences to replace them by noexcept(false). If the number of occurrences is too high to make a patch viable, then similarly to mothur, the advice from Aaron M. Ucko applies: informing upstream and temporarily getting back to C++14 is probably the saner course of action.
Maketh sense, thanks for explaining. I guess the main problem here will be with dead upstreams, which have a number of these occurances, looks like we need to propagate hacks (probably for a very long time) for these cases. For disulfinder, the patch does not look as simple as changing 3 lines, so I'm admittedly tempted to force c++14 here :-P Nilesh
Attachment:
OpenPGP_signature
Description: OpenPGP digital signature