Right, I changed my mind: mingw_progs as the option name is much
more comprehensible.
Regards Henri
On 1/7/26 19:19, J.H. vd Water via Cygwin-patches wrote:
> On 1/7/26 13:51, Jon Turney wrote:
> [snip]
>> Thanks for looking into this. And thanks for providing a patch.
>>
>> Yes, this is all messed up!
>>
>> This all looks fine.
>>
>> I think changing the option name is the best approach, as it's just
>> completely unclear what it means either way around, and would be
>> even more confusing if we suddenly inverted its meaning.
>>
>> But how do you feel about reverting to the previous name '--without-
>> mingw-progs' (this exactly matches what the logic does now, since
>> the extra things that '--with(out?)-cross_bootstrap' was omitting
>> have been unconditionally removed in the meantime)?
> ? The option name was previously: mingw_progs (well, it was: mingw-progs).
>
> Yes, the switch in the help message was: --without-mingw-progs
>
> Omitting?
>
> I guess you refer to the mingw-runtime (was: mingw-crt), a dependency that
> has been moved to the compiler ...
>
> Of course, "mingw_progs" can be chosen as the option name ...
>
> BUT ... There is a reason why the macro is named AC_ARG_WITH (and why it is
> not named AC_ARG_WITHOUT)
>
> - switch --without-FOO results in with_FOO = "no", No matter the word that
> is substituted for FOO
>
> - one cannot specify an argument to the --without-FOO switch; it will result
> in an error (abort by configure) if one does
>
> - the action-if-not-given must be with_FOO = "yes" i.s.o. "no" if the choice
> is for "mingw_progs" as the option name, because "by default" (i.e. if no
> switch is specified), the process should hunt for the MinGW Toolchain.
>
> As result, the choice for "mingw_progs" will require not testing for "! yes",
> but for "! no"; because the value assigned to with_FOO can be different from
> either "yes" or "no" (as result of a typo).
>
> if with_FOO != "no"
> then
> Hunt for mingw # also in case of a typo
> fi
>
> (and that does also apply to AM_CONDITIONAL macro that follows this test)
>
> And personally, I doubt that I have the energy to do the whole process of
> fighting against "git" again.
>
> And that is why I prefer "skip_mingw" as the option name ... Sorry.
>
> Note: both winsup/{testsuite,utils}/Makefile.am must be modified as well.
>
> So, go ahead if your choice is for "mingw_progs" :-)
>
> Regards Henri
>
> ---
> The choice is therefore between:
>
> AC_ARG_WITH([skip_mingw],
> [AS_HELP_STRING([--with-skip-mingw],
> [do not build programs using the MinGW toolchain])],
> [],
> [with_skip_mingw=no])
>
> if test "x$with_skip_mingw" != "xyes"; then
> Hunt for the MinGW Toolchain
> fi
> AM_CONDITIONAL(SKIP_MINGW, [test "x$with_skip_mingw" != "xyes"])
>
> and ...
>
> AC_ARG_WITH([mingw_progs],
> [AS_HELP_STRING([--without-mingw-progs],
> [do not build programs using the MinGW toolchain])],
> [],
> [with_skip_mingw=yes])
>
> if test "x$with_skip_mingw" != "xno"; then
> Hunt for the MinGW Toolchain
> fi
> AM_CONDITIONAL(SKIP_MINGW, [test "x$with_skip_mingw" != "xno"])
>
> # --with-FOO => with_ := yes Hunt for the MinGW Toolchain.
> # --with-FOO=yes => with_ := yes Hunt for the MinGW Toolchain.
> # --without-FOO => with_ := no do NOT hunt for the MinGW Toolchain.
> # --with-FOO=no => with_ := no do NOT hunt for the MinGW Toolchain.
> # --with-FOO=none => with_ := none Hunt for the MinGW Toolchain.
> # (none = anything else than yes or no)
>
> =====
