Refactoring patches

[José Abílio Matos]

On Tuesday, October 13, 2020 10:41:28 PM WEST Pavel Sanda wrote:
> And then there is somewhat separate issue about generic system headers which
> sometimes land in the code because someone brave enough tried to run it on
> solaris or haiku or god knows what. Here I would be more conservative, and
> would not just blindly delete those because I think
> i) there will be few of them so one could possibly check git history how it
> appeared
> ii) the speedup gain is negligible compared to problems we could
> cause elsewhere
> 
> Pavel

You are right in your analysis (also for the part that I did not quoted here).
The problem is, of course, the include system that was inherited from C and 
the C++ modules that are not here yet. Although the modules are part of C++20 
the STL are not yet ported to use them, we will need to wait for C++23 and 
then that the support is universal enough so that we can require it. :-)

In the meantime (that can be long) I suggest that every system header that is 
included because of some quirk in any system should be documented in place.
It is not easy or even reasonable to require git blame or to read the log 
messages to understand why the included was added in the first place.

-- 
José Abílio
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lyx.org/pipermail/lyx-devel/attachments/20201014/bb1a8b50/attachment.html>