Re: [hatari-devel] Hatari 2.6.1 startup failure on Windows |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/hatari-devel Archives
]
- To: hatari-devel@xxxxxxxxxxxxxxxxxxx, Christian Zietz <czietz@xxxxxxx>
- Subject: Re: [hatari-devel] Hatari 2.6.1 startup failure on Windows
- From: Nicolas Pomarède <npomarede@xxxxxxxxxxxx>
- Date: Sat, 16 Aug 2025 19:18:48 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/simple; d=corp.free.fr; s=rcs; t=1755364733; bh=PEWKqeEEJyx0VCED4olnEXTbXdP9gCuZjAcEZ8Rm6OM=; h=Date:Subject:To:References:From:In-Reply-To:From; b=oBSE/w58WmJnJobGvbpFUS9cBF+qgS6Pm7Nz7DHcfp5xlnXcecziWp2kxSCU+2vzu YNj+gdB4cWGnuDsKUS9MtdF9kaVThLDWZ76ID8GIvomq4Hfq9+L9jetlR7rZkQQXvo IanNZ8+69jM1KGXwhWLkqX6+6sPp5mqW7v2ZsJIT8Vz4iLV4U0O1yhugFRvMSEonFt oEapYu3+0hcl8jwCF0ARjS//wKpa5mcgcomVO5te2EcfaV5BarCKxMSDvp65csDPqw qnv6M01TUN6O8jc0wcBfPIEZbC7KTjck5AmocuxG4qTvyKJ+KxaD8dHAfYfwVvLLE9 4GmXtm7Jhoa8Q==
Le 16/08/2025 à 14:16, Christian Zietz a écrit :
Hello,
I suppose you already saw: https://www.atari-forum.com/viewtopic.php?
p=484369#p484369. Hatari 2.6.1 release fails to start on Windows.
After looking into it, I presume that the headers (devel package) for
winpthreads was updated to now include nanosleep64 instead of nanosleep.
But the libwinpthread-1.dll shipped with the release is older; it's the
same that shipped with 2.6.0, and only exports nanosleep. If I replace
said DLL by the current binary from https://packages.msys2.org/packages/
mingw-w64-x86_64-libwinpthread, Hatari launches without error message.
The release archive should be updated accordingly. Perhaps the archive
should also be renamed, to make clear that its contents has changed.
Regards
Christian
PS: Interestingly, the pipeline build is not affected. Therefore, the
build environments for pipeline and release must be different. Perhaps
it's worth discussing whether Hatari should use its CI/CD pipeline to
build releases, like EmuTOS does. In that way you get some test coverage
of the build environment when people use pipeline builds.
Hi
thanks for the report, as mentionned on atari-forum.com this is now
fixed and was due to an unnoticed package update on my linux pc + not
correctly testing with "wine"
Nicolas