+1.

-- Gaby


________________________________
From: Ext <ext-bounces_at_[hidden]> on behalf of Steve Downey via Ext <ext_at_[hidden]>
Sent: Saturday, April 16, 2022, 1:57 PM
To: Evolution Working Group mailing list <ext_at_[hidden]>
Cc: Steve Downey <sdowney_at_[hidden]>; Daniel Ruoso <daniel_at_[hidden]>; ISO C++ Tooling Study Group <sg15_at_[hidden]>
Subject: Re: [isocpp-ext] [SG15] Can we expect that all C++ source files can have the same suffix?



On Sat, Apr 16, 2022 at 2:47 PM René Ferdinand Rivera Morell via Ext <ext_at_[hidden]<mailto:ext_at_[hidden]>> wrote:
On Sat, Apr 16, 2022 at 1:41 PM Daniel Ruoso via SG15 <sg15_at_[hidden]<mailto:sg15_at_[hidden]>> wrote:
I, personally, think it's time to get over the expectation that it's realistic to use c++ without a build system.

Especially in teaching context, we should prepare folks to the real world, which necessarily involves a build system, instead of creating this anachronistic requirements.

It is important for programmers, especially for C++, to understand the entire pipeline that they use. Relying on tools, like build systems and IDEs, hiding the knowledge makes for inefficient programmers. Learning that pipeline starts at invoking the compiler directly.

But even as a build system engineer, I turn off verbose output of build commands all the time. In my production builds it's often a K or two of flags and options, and not terribly meaningful unless I'm debugging a weird compilation error and need to get the preprocessed source. I think C++ engineers should have a reading knowledge of assembler, but that's not a good place to start when learning. The best path doesn't really have to trace the entire history of tool development.

Even python treats having `setup.py` for build and install as a 102 level course.