Re: smlnj11072 removal proposal

[Thomas Klausner <wiz%NetBSD.org@localhost>]

On Sun, Mar 23, 2025 at 10:48:18PM +0000, David Holland wrote:
> On Tue, Mar 18, 2025 at 11:53:30AM +0100, Thomas Klausner wrote:
>  > I propose removing lang/smlnj11072. It is an older version of
>  > lang/smlnj, which doesn't build in any bulk builds I can find.
>  > 
>  > https://releng.netbsd.org/bulktracker/lang/smlnj11072
>  > 
>  > I looked at the CVS history, and it was added in 2016 because of
>  > lang/smlnj's lack of i386 support. However, lang/smlnj seems to
>  > support i386 fine nowadays, see
>  > 
>  > https://releng.netbsd.org/bulktracker/lang/smlnj
>  > 
>  > So, let's remove lang/smlnj11072.
> 
> Go ahead, it was only there because of glaring build problems in the
> newer version.

Thanks, done.

> At the time, it often took multiple tries to get a successful build of
> _either_ version (vs. segfaulting) but as I recall the success rate of
> the new version on i386 was identically 0.
> 
> I see that lang/smlnj is a considerably newer version; I take it that
> means it actually has an upstream again, so we could get the time_t
> patches upstreamed?

Yes, it has an active github project. The version we are using is the
'legacy' version (a 'development' version also exists), a slightly
newer release is out (110.99.7.1), and you can file bug reports/PRs at
github: https://github.com/smlnj/legacy
 Thomas