[Nasm-devel] Two comments about revision g1df7263ae

[C. M. <pu...@38...>]

Hello,

I reviewed the revision at
https://repo.or.cz/nasm.git/commitdiff/1df7263ae937ac11abb2c6938b8891745af91ce6


This change seems to be wrong:

@@ -1575,10 +1570,6 @@ static void assemble_file(const char *fname,
struct strlist *depend_list)

         preproc->cleanup_pass();

-        /* Don't output further messages if we are dead anyway */
-        if (terminate_after_phase)
-            break;
-
         if (global_offset_changed) {
             switch (pass_type()) {
             case PASS_OPT:


And in the following:

+/* Pop the warning status off the warning stack */
+void pop_warnings(void)
+{
+       struct warning_stack *ws = warning_stack;
+
+       memcpy(warning_state, ws->state, sizeof warning_state);
+       if (!ws->next) {
+               /*!
+                *!warn-stack-empty [on] warning stack empty
+                *!  a [WARNING POP] directive was executed when
+                *!  the warning stack is empty. This is treated
+                *!  as a [WARNING *all] directive.
+                */
+               nasm_warn(WARN_WARN_STACK_EMPTY, "warning stack empty");
+       } else {
+               warning_stack = ws->next;
+               nasm_free(ws);
+       }

... It seems that WARN_WARN_STACK_EMPTY is issued depending on the *new*
state of warning_state, meaning that warn-stack-empty can only be
disabled or enabled from the command line. I think that it'd be more
useful to issue the warning before applying the original warning_state
content, so that the warning warn-stack-empty could be disabled by the
input file without using the command line.

Regards,
ecm