Re: string-backed FILEs mess

[Rich Felker <dalias@libc.org>]

On Wed, Sep 12, 2018 at 10:55:45AM -0500, A. Wilcox wrote:
> On 09/12/18 10:09, Markus Wichmann wrote:
> > I don't know how important the glibc-compat still is for people. I don't
> > need it, but then, I have glibc available if necessary. And with Steam
> > moving to Linux, closed-source software is certainly going to be on the
> > rise for Linux/PC (i.e. Linux/i386 and Linux/amd64). Which is going to
> > be a problem for musl-based distributions, if they want to support that.
> 
> 
> Genuinely, I used it every day I had an x86_64 box.  Spotify was the
> main reason (the Qt 4 app; NOT the CEF/chrome-based piece of ****).  Of
> course, now that I trashed my last x86 and am surrounded by better cores
> (almost exclusively PowerPC, with some ARM and MIPS for good measure),
> glibc ABI compat is almost worthless.
> 
> However, that's what https://code.foxkit.us/adelie/gcompat/ is for
> anyway.  Since it uses LD_PRELOAD, we could probably overwrite the
> entirety of stdio with something compatible with glibc if necessary.

As described in the post Markus was replying to, this is a non-issue.
If the design were going to break compat it would just put
must-be-zero fields over top of the old glibc positions so that they'd
always force an actual function call. This would actually reduce the
ABI surface down from actual field values and their interpretation to
a mere contract that certain bytes be all-zeros.

Rich