Re: Increasing Qt dependency in master

Nah, I agree that's pretty bad. Requiring a minimum stable release is fine.

That said, I would be fine with not supporting the weird 5.12.0 case
in the code at all too. Whoever is using 5.12.0 isn't short of
irresponsible.

Aleix

On Wed, Oct 28, 2020 at 6:16 PM Igor Kushnir <[email protected]> wrote:
>
> On 2020-10-28 18:54, Aleix Pol wrote:
>
> On Mon, Oct 26, 2020 at 11:28 AM Igor Kushnir <[email protected]> wrote:
>
> On 2020-10-26 12:09, Milian Wolff wrote:
>
> On Montag, 26. Oktober 2020 09:30:30 CET Igor Kushnir wrote:
>
> If we do make the leap to 5.12, I'd like it to be straight to 5.12.1
> rather than 5.12.0 to ease porting from QRegExp::Wildcard to
> QRegularExpression::wildcardToRegularExpression(). The behavior of this
> function was changed in Qt 5.12.1 as it started anchoring its return
> value to fix QTBUG-72539.
>
> Personally I'm fine with that.
>
> What distros would we lose by this move?
>
> Cheers
>
> By requiring 5.12.0 we'd lose Debian stable [buster], old Ubuntu LTS
> [18.04] and old OpenSUSE Leap [15.1]. By requiring 5.12.1 we'd not lose
> any extra popular distro version in addition to these 3 releases, I think.
>
> Igor
>
> +1 on changing to 5.12.
>
> @Igor: We generally never depend on specific minor versions. In fact,
> I'd expect any maintained distribution to be shipping the last minor
> release, if they don't they're doing a big disservice to their users.
>
> Aleix
>
> Aleix, would you rather set(QT_MIN_VERSION "5.12.0"), but assume Qt 5.12.1+ to avoid conditional compilation like this?
>
> Igor