[XS] : "snprintf" portability options

syphilis has asked for the wisdom of the Perl Monks concerning the following question:

Hi,

In Cpanel-JSON-XS-4.38, the XS.xs file contains a few lines like:

snprintf (enc->cur, IVUV_MAXCHARS, "%" UVuf, uv)
[download]

which is fine on *nix systems but on Windows builds of perl, whenever IVSIZE is 8, it warns:

XS.xs:2449:66: warning: format '%u' expects argument of type 'unsigned
+ int', but argument 4 has type 'long long unsigned int' [-Wformat=]
[download]

I can fix that for those Windows perls by replacing the occurrence of "%" with "%llu" but I doubt the portability of that fix. (I guess I could also add separate renditions for #if defined(WIN32) && IVSIZE == 8 .... yuk.)

What's the best portable way of dealing with this ? ... and where are the options for this formatting style documented ?
I can never remember where it that documentation is, and I can never find it when I need it :-(

Cheers,
Rob

Comment on [XS] : "snprintf" portability options Select or Download Code

Replies are listed 'Best First'.
Re: [XS] : "snprintf" portability options by tonyc (Friar) on Sep 15, 2024 at 22:32 UTC
I suspect you're running into gcc -Wformat not supporting the Windows specific "%I64u" format. This is why win32/GNNUmakefile adds -Wno-format after -Wall and -Wextra.	[reply]
Re^2: [XS] : "snprintf" portability options by syphilis (Archbishop) on Sep 16, 2024 at 00:53 UTC
This is why win32/GNNUmakefile adds -Wno-format after -Wall and -Wextra And the Cpanel-JSON-XS-4.38 Makefile.PL then oh-so-thoughtfully re-introduces the issue by appending "-Wall -Wextra -W" .... I'll submit a PR for that Makefile.PL, along with a PR to XS.xs that allows Cpanel::JSON::XS to build on Windows quadmath perls (and also silences some other warnings). Thanks tonyc - I remember that issue, now. At one stage yesterday it crossed my mind that I don't usually encounter `[-Wformat=]` warnings ... but I still failed to make the connection :-( Cheers, Rob	[reply] [d/l]
Re^3: [XS] : "snprintf" portability options by ikegami (Patriarch) on Sep 16, 2024 at 10:27 UTC
oh-so-thoughtfully re-introduces the issue No it doesn't. The issue is there with or without the `-Wformat`. Your sarcasm is misplaced since the warning isn't spurious. The issue is that the wrong format is being used. This is a real error that needs fixing.	[reply] [d/l]
Re^4: [XS] : "snprintf" portability options by tonyc (Friar) on Sep 16, 2024 at 22:56 UTC
Re^5: [XS] : "snprintf" portability options by NERDVANA (Priest) on Sep 17, 2024 at 01:34 UTC
Some notes below your chosen depth have not been shown here
Re^5: [XS] : "snprintf" portability options by ikegami (Patriarch) on Sep 17, 2024 at 10:25 UTC
Re^4: [XS] : "snprintf" portability options by etj (Priest) on Sep 16, 2024 at 13:09 UTC
Re^5: [XS] : "snprintf" portability options by hippo (Archbishop) on Sep 17, 2024 at 12:48 UTC
Re^5: [XS] : "snprintf" portability options by jdporter (Paladin) on Sep 17, 2024 at 15:34 UTC
Some notes below your chosen depth have not been shown here
Re^5: [XS] : "snprintf" portability options (bugfix) by LanX (Saint) on Sep 17, 2024 at 11:23 UTC
Some notes below your chosen depth have not been shown here
Re: [XS] : "snprintf" portability options by etj (Priest) on Sep 15, 2024 at 11:12 UTC
Surely the `UVuf` is entirely supposed to give the right format to match the type of UV. If it's not doing so, that's a bug in Perl, and should be fixed there?	[reply] [d/l]
Re^2: [XS] : "snprintf" portability options by syphilis (Archbishop) on Sep 15, 2024 at 13:59 UTC
Surely the UVuf is entirely supposed to give the right format to match the type of UV. Sounds reasonable and I can't re-create the warnings in an Inline::C script - there might be more to this than I initially realized. Here's a bit more detail regarding one of those warnings: `XS.xs: In function 'encode_sv': XS.xs:2449:66: warning: format '%u' expects argument of type 'unsigned + int', but argument 4 has type 'long long unsigned int' [-Wformat=] 2449 \| enc->cur += snprintf (enc->cur, IVUV_MAXCHAR +S, "%" UVuf, UV_MAX); \| + ^~~` [download] (It's getting late over here.) Cheers, Rob	[reply] [d/l]
Re^3: [XS] : "snprintf" portability options by NERDVANA (Priest) on Sep 16, 2024 at 03:24 UTC
I hear a lot about new versions of gcc making it a fatal error to declare "void main", which tends to break short test programs during configure scripts and results in tests coming out false when they should return true. They can be hard to catch if a test is false on a majority of systems anyway. Maybe perl is affected?	[reply]