Hello, On Fri 04 Aug 2023 at 09:08pm +02, Aymeric Agon-Rambosson wrote:
> That is not entirely true. The C variable > Vinhibit_automatic_native_compilation, corresponding to elisp variable > inhibit-automatic-native-compilation, is effectively used to disable native > compilation of loaded elc files. > > The patch remains useless nonetheless : this (partial) disabling of native > compilation is redundant with the one already provided by variable > native-comp-jit-compilation. > > In fact, if you look at this part of the patch : > > diff --git a/src/comp.c b/src/comp.c > index 59c9e9619a3..06ddb956bb5 100644 > --- a/src/comp.c > +++ b/src/comp.c > @@ -5179,6 +5179,7 @@ maybe_defer_native_compilation (Lisp_Object > function_name, > return; > > if (!native_comp_jit_compilation > + || !NILP (Vinhibit_native_compilation) > || noninteractive > || !NILP (Vpurify_flag) > || !COMPILEDP (definition) > > > You will notice that the first two boolean conditions are redundant : > native_comp_jit_compilation already does what Vinhibit_native_compilation > wants to do, albeit with an inverted meaning. But the patch allows it to be done using an env var. There is no env var corresponding to the new Vnative_comp_jit_compilation. And the env var is needed so that we can be sure it's turned off early enough. So, while it would be tidier to rework the patch to avoid introducing a second variable, its functionality is still useful. -- Sean Whitton
signature.asc
Description: PGP signature
