Fix build on 32 bit arches

[david-geiger]

No description provided.

[david-geiger]

Mageia Cauldron + GCC14:

On 32bit arches without this patch we got this following build issue:

In file included from /home/iurt/rpmbuild/BUILD/gnucash-5.6/libgnucash/engine/qof.h:74, from /home/iurt/rpmbuild/BUILD/gnucash-5.6/libgnucash/engine/gnc-engine.h:41, from /home/iurt/rpmbuild/BUILD/gnucash-5.6/libgnucash/engine/gnc-commodity.h:55, from /home/iurt/rpmbuild/BUILD/gnucash-5.6/libgnucash/engine/Transaction.h:93, from /home/iurt/rpmbuild/BUILD/gnucash-5.6/gnucash/import-export/import-main-matcher.h:40, from /home/iurt/rpmbuild/BUILD/gnucash-5.6/gnucash/import-export/import-main-matcher.cpp:45: /home/iurt/rpmbuild/BUILD/gnucash-5.6/gnucash/import-export/import-main-matcher.cpp: In function 'void gnc_gen_trans_assign_transfer_account_to_selection_cb(GtkMenuItem*, GNCImportMainMatcher*)': /home/iurt/rpmbuild/BUILD/gnucash-5.6/libgnucash/engine/qoflog.h:267:16: error: format '%ld' expects argument of type 'long int', but argument 5 has type 'std::vector<std::unique_ptr<_GtkTreeRowReference, TreeRowRefDestructor> >::size_type' {aka 'unsigned int'} [-Werror=format=] 267 | "[%s] " format, PRETTY_FUNC_NAME , ## args); \ /home/iurt/rpmbuild/BUILD/gnucash-5.6/gnucash/import-export/import-main-matcher.cpp:850:5: note: in expansion of macro 'DEBUG' 850 | DEBUG("Rows in selection = %ld", selected_refs.size()); | ^~~~~ /home/iurt/rpmbuild/BUILD/gnucash-5.6/gnucash/import-export/import-main-matcher.cpp:850:34: note: format string is defined here 850 | DEBUG("Rows in selection = %ld", selected_refs.size()); | ~~^ | | | long int | %d cc1plus: all warnings being treated as errors

[christopherlam]

Does %zu work?

[jralls]

Does %zu work?

See #1904

[christopherlam]

Ok. Would it be an idea to including 32-bit builds on CI?

[jralls]

Ok. Would it be an idea to including 32-bit builds on CI?

We do sort of with the Windows nightlies, but apparently MSYS2's version of gcc didn't flag up this parameter mismatch.

[madpilot78]

Does %zu work?

See #1904

I have tested my patch with %zu on FreeBSD, and it works fine there for compiling. I am not sure how to get that debug code to actually execute.

BTW I'm using my patch in the FreeBSD ports tree at present: freebsd/freebsd-ports@3abcd80

[jralls]

To execute the debug statement start GnuCash with --log gnc.import.main-matcher=debug and import a file.

%zu is more correct than casting the size_t to long int so we'll go with #1904.