C++ Logo


Advanced search

Re: [SG15] module source suffixes

From: Matthew Woehlke <mwoehlke.floss_at_[hidden]>
Date: Wed, 28 Aug 2019 12:58:43 -0400
On 28/08/2019 00.55, Gabriel Dos Reis wrote:
> [...]
> | Someone is wrong here.
> Yes 😊
> You are trying to prescribe here how VS IDE should work internally.

Please take a step back from "trying to protect your baby" and re-read
my previous message, noting in particular that I did not state *which*
of the two competing possibilities I believe to be correct. In fact, I
am especially concerned because the VS claim may be *right*.

The VS devs claim that VS cannot be better implemented; that it is
impossible to achieve reasonable performance without knowing *from the
file name* if a source exports stuff.

What I *don't* see is how this can be unique to VS. If true, it should
be true for *every* IDE, and very probably for build systems as well. If
not true, it implies that VS could be implemented in a way that wouldn't
have this problem, and that therefore it doesn't make sense to codify
such a restriction in the standard.

Other involved parties don't believe this to be an issue. So... is VS
aware of something that the rest of us haven't noticed yet? (Entirely
possible; VS has a large development team that is strongly motivated
toward early adoption compared to other IDE developers.) Or is VS trying
to dictate an unnecessary restriction on how modules work?

*I don't know*. I would like to know. I would like for us to have a
better handle than I feel like we do currently on which of these is
correct, so that we are making a properly informed decision on this topic.

*If*, in fact, there is no technical reason why an IDE cannot be made
efficient without the standard mandating a special file name convention
for sources with exports, then yes, I am *perfectly* happy for the
standard to dictate that implementations must meet certain "quality"
standards. We've been doing that for decades.

All that said, I'm far from convinced that we (WG21) should be dictating
this sort of thing even if it *is* a legitimate problem. It feels more
like a QoI issue. Nothing is stopping VS from making its own
recommendations, or WG21 from adopting a "wait and see" attitude to
allow more real world experience before codifying something.


Received on 2019-08-28 12:00:48