Got a chance to test this yesterday. Your change fixed it indeed.
Regards,
Geert
Op donderdag 2 maart 2023 21:36:00 CET schreef John Ralls:
> Found and fixed it. It turns out that string->number isn't locale-sensitive,
> std::to_string is. I replaced it with ostringstream and pushed.
>
>
Found and fixed it. It turns out that string->number isn't locale-sensitive,
std::to_string is.
I replaced it with ostringstream and pushed.
Regards,
John Ralls
> On Mar 2, 2023, at 9:16 AM, john wrote:
>
> Geert,
>
> Aha! That would certainly explain why I couldn't reproduce it. That
Geert,
Aha! That would certainly explain why I couldn't reproduce it. That change is
from
https://github.com/Gnucash/gnucash/commit/408b5ec2165fb43bef18b80ff44ba4d2f1b67f5f.
Yes, no doubt string->number expects the number to be represented in the
current locale, so it wants "42,0" and
I still haven't figured out exactly where it goes wrong, but I did discover
it's a locale related
issue. For some reason this particular test has become locale sensitive.
If I run the test as follows:
LC_ALL=C ctest -V -R test-gnc-option-scheme-output
it passes. If however I do this:
Geert,
I can't replicate that, tried on Debian 10, Debian Unstable, and Arch. I'm
installing a Fedora 37 VM but I'm out of time for today and won't get to the
point of a test build of GnuCash.
Regards,
John Ralls
> On Feb 28, 2023, at 2:07 PM, Geert Janssens
> wrote:
>
> Hi,
>
> With the