Hacker News Reader

Comment by IAmLiterallyAB | Hacker News Reader

The progress is frustratingly slow. My understanding is GCC and Clang still haven't finished implementing them fully. Last I read they are still making significant changes, for example moving to the strong ownership model for symbols. Once they're done, hopefully the build system and IDE support will follow quickly. MSVC seems to be in much better shape.

View on HN

The progress is mostly unfunded. It's not that the work is especially slow (maybe it is a little?). Mostly it's that barely anyone is working on it.

This is especially frustrating for organizations ostensibly paying vendors for high quality compilers.

It's frustrating because consider how many hours are lost to compile times over the whole industry.

I have no idea how many people are slopping C++ code. So lets just calculate it per 1000. Assume modules save an average of 10 minutes a day.

1000 X 250 days/year X 0.167 hours/day -> 41,666 hours a year per thousand coders. Or 21 man years.

Yeah you'd think it'd be worth funding heavily.

When you use a good build system like Bazel modules do not save 10 minutes a day. The time saving is negligible. That's why large C++ shops do not care about modules and only volunteers are working on them.

Not everyone uses bazel. I would wager that a vanishingly small amount of people use bazel. If people using other things was enough to not let things be standardized we wouldn't have asio, ranges, fmt. Precompiled headers also exist, so why would we standardise modules?

Reply on news.ycombinator.com