Re: [PATCH 0/11] renaming argv_array

From: Jeff King <peff@peff.net>
Date: Aug 12, 2020, 15:06 UTC
Message-ID: <20200812150611.GA33189@coredump.intra.peff.net>
In-Reply-To: <82991f30-fe37-d6d2-ffd5-8b0878f46c83@web.de>

On Tue, Aug 11, 2020 at 06:08:22PM +0200, René Scharfe wrote:

Show 13 quoted lines

> > This has bugged me for a while, so I decided to finally fix it. It
> > wasn't _too_ painful, though I'm sure there will be a little fallout
> > with topics in flight.
> 
> Just as this landed in master now, https://lobste.rs/ decided to link to
> http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2493.pdf, which is a
> paper about reserved identifiers in C.  It contains a nice overview.
> 
> Anyway, 7.31 of C11 says: "All external names described below are
> reserved no matter what headers are included by the program."  And
> 7.31.13 goes on: "Function names that begin with str, mem, or wcs and a
> lowercase letter may be added to the declarations in the <string.h>
> header."  So the names of the strvec functions are reserved.

It was the same in C99. I think this is one of those cases where we need to worry less about what the standard says and more about what the real world does. As long as we're ready for C25 or whatever to add "strvec" and we accept that we'll need to change the name then, I'm not particularly concerned. A compiler that starts warning about "str" functions in the meantime would be impractical I think (forget strbuf, stuff like strip_extension() would be illegal).

Show 5 quoted lines

> Also how about using Coccinelle and patience to reduce the impact of
> such a change next time?  I.e. adding the new thing, providing a
> semantic patch for converting old code, waiting a reasonable amount of
> time after the last conversion was necessary and then removing the
> old thing.

So I almost sent a rant about Coccinelle along with this series. :)

Debian unstable now ships coccinelle 1.0.8, and it's unbelievably slow compared to 1.0.4. Running "make coccicheck" is currently at 80 minutes of CPU time running each script in parallel, with none of them down. They're also all consuming 6GB of RAM each, so I'm killing them all.

I got somewhere similar when I was working on this series, got fed up, and then just did the whole thing with grep, which was easier. I'm open to the idea that a slower transition might have helped topics in flight catch up, but it also would have prolonged the pain. So I dunno.

-Peff