bug#12215: CSET is unnecessarily confusing

[Paul Eggert]

>> currently the people who are actually doing those experiments
>> (Dmitry, Tom) don't need these macros and don't particularly want them.
> 
> AFAIK that's true of BVAR but not of setter macros (which Dmitry does need).

Dmitry suggested in <http://bugs.gnu.org/12215#49> to remove the setters
from the trunk, and that he try them out in the GC branch.
Although Dmitry wrote that he's quite comfortable with setter functions
that are currently in the trunk, the setters could be also macros using
the syntax that you're thinking of.  Either way, Dmitry could gain
experience with setters in the GC branch, and we wouldn't have to decide
now how setters should work in the trunk.
 
>> And Chong is calling BVAR "horrible".
> 
> Yes, tho the issue is the use of field names in syntactic contexts where
> they can be confused for variable names.

That's the main issue, I guess.  But BVAR also makes code harder to read.