[lpyu001]

Preserve the default enabled safety configuration while correctly handling enable, disable, and invalid values for both safety options.

• Issue: configure mishandles the default and --disable-* forms of the safety options #156764

[lpyu001]

Hi, could you take a look at this PR when you get a chance @corona10

[read-the-docs-community]

Documentation build overview

📚 cpython-previews | 🛠️ Build #35569044 | 📁 Comparing b3f6a9e against main (a6b4aa6)

  🔍 Preview build  

135 files changed · + 1 added · ± 134 modified

+ Added

• c-api/pyexpat.html

± Modified

• glossary.html
• license.html
• py-modindex.html
• builtins/exceptions.html
• builtins/functions.html
• builtins/stdtypes.html
• c-api/bool.html
• c-api/buffer.html
• c-api/bytes.html
• c-api/code.html
• and 124 more...

[lpyu001]

I've updated this PR. Both options now stay disabled by default, as decided in gh-124064.

• configure.ac / configure: use $enableval, so --disable-safety and --disable-slower-safety no longer enable the options. Invalid values are rejected.
• Doc/using/configure.rst and Doc/whatsnew/3.14.rst: document --enable-safety (off by default) instead of the old --disable-safety.

This affects 3.14 and 3.15 as well, so it should be backported.

[lpyu001]

cc @AA-Turner

[maaz7409]

Looks Good, the --disable-* fix works in my Linux checks. Also, would it make sense to keep accepting values such as =1, =true and an empty =? These currently enable the flags, so rejecting them could break existing build scripts after a backport.

[maaz7409]

Would treating any value other than no as enabled work here? This would keep --enable-safety=1, =true and an empty = working while honoring --disable-safety:

Suggested change

	[AS_CASE([$enableval],
	[yes|no], [enable_safety=$enableval],
	[AC_MSG_ERROR([invalid --enable-safety option: expected "yes" or "no"])])],
	[AS_VAR_IF([enable_safety], [no], [], [enable_safety=yes])],

[maaz7409]

Perhaps the same approach could work for --enable-slower-safety too? If these suggestions are used, configure would also need regenerating.

Suggested change

	[AS_CASE([$enableval],
	[yes|no], [enable_slower_safety=$enableval],
	[AC_MSG_ERROR([invalid --enable-slower-safety option: expected "yes" or "no"])])],
	[AS_VAR_IF([enable_slower_safety], [no], [], [enable_slower_safety=yes])],

[lpyu001]

Thank you for the great suggestion! I'll update my code accordingly.