Re: 10.1 make oddity

[Roland Illig <roland.illig%gmx.de@localhost>]

Am 09.06.2026 um 11:44 schrieb Edgar Fuß:
>> The "$;" is then not interpreted as an expression with variable name ";"
> Which, in fact, is correct. The $ is a RE anchor, the ; the RE terminator.

There are many possible interpretations of what is "correct", yours is
one of them. The manual page at
https://man.netbsd.org/make.1#FILE%20DEPENDENCY%20SPECIFICATIONS doesn't
even mention that expressions are allowed in dependency lines, which
could mean that all bets are off at this point.

POSIX says at
https://pubs.opengroup.org/onlinepubs/9799919799/utilities/make.html#tag_20_76_13_05
that "Macros can appear anywhere in the makefile" and that in target
lines, they are "evaluated when the target line is read", which still
gives plenty of room as to when exactly they are expanded and thus how
the semicolon is to be found. It could be searched for before expanding
the macros or after. The current implementation first searches for ";",
then expands the line, and finally searches for the ":" between targets
and sources. This means that a separator ";" cannot be generated by an
expression, but a separator ":" can. This asymmetry may be a corner case
and is probably not exploited in practice, but it is there.

I fixed the code for finding the semicolon so that it conforms to your
interpretation, assuming that many other people expect the same.
Splitting the line in the middle of an expression was so surprising that
I had to step through the code to find the accurate explanation.

Roland