"Thomas Frohwein" <tfrohw...@fastmail.com> wrote:
>
> For the record, I'm against dynamic core and against disabling the
> splash. I'm indifferent regarding ne2000.
Since the latter seems to be the least controversial, perhaps meshould
trim the patch down to that first...? jsg@, you're maintainer, please
speak up :)
> The ideal future state would be removing W|X from those remaining
> ports.
In the ideal future ports wouldn't be needed and all third-party
software would fit in perfectly with base ;) But yeah.
> Therefore, the burden of proof should lie with anyone wanting
> to go in the opposite direction. You want dynamic core? Convince us
> that the benefit is worth going _against_ our security mitigations. You
> should seriously consider showing numbers or comparison videos to make
> your case.
Uhm, as me's said, me's not *forcing* anyone to run the dyncore... me's
not even proposing it as a default. Just an option.
If this were in base, mewouldn't bother proposing it, but hey, this is
ports, it's full of junk that people need despite it being so.
> Even then I'm likely not going to think it's worth it. There is a fork
> of dosbox called dosbox-x [1] that has seen continuous improvement and
> regular releases over the recent years.
Mehasn't been monitoring dosbox development much, due to the lack of
Code quality and indeed plain English quality... mewas thus unaware of
the fork and will investigate, thanks for bringing it me attention.
> The project is complex and a
> little messy, which is why I haven't sent it to ports@ yet. I had an
> SDL1-based version that worked very well about a year ago. There was a
> noticeable performance advantage in the game Tie Fighter over then
> dosbox 0.74, with the normal core. That to me was good enough to lose
> any interest in the dynamic core. It even runs Windows 98 acceptably,
> but 3D acceleration isn't fully there in Win98.
Medoesn't know that game, but meexperience is that some games really do
try to be too clever with optimizations, turning them into
pessimizations within an emulator of any kind. The dynamic core somewhat
(sometimes greatly) alleviates that.
> The port has since seen a few updates. I just built the most recent one
> with SDL2, but there are several bugs that I'd like to address before
> it's ready.
>
> I would recommend to check out dosbox-x and how it performs with its
> normal core before looking into dynamic core and the associated
> reduction in mitigations.
IME running dosbox in general is not so much putting the door ajar as it
is removing it from its frame entirely and junking it.
>>> A better way to spend time on dosbox would be to investigate ways to
>>> improve speed without sacrificing basic security protections.
>
> dosbox-x may offer this; however I haven't tried most recent dosbox
> 0.74-2 yet. I should soon have a version to share for testing.
Well, if this "dosbox-x" does the job, we of course don't need the
dyncore. Just don't make it drag in gtk+? and python (or similar
horrors), please...
> I think upstream made it clear enough that the splash should remain
> part of the application. Imagine the mess if we started adding patches
> to ports for anything that someone might consider "convenient".
It brings the functionality of the port closer to base, doesn't it?
Honestly, I don't find "you're allowed to modify everything *except*
that bit", when "that bit" gets in the way, much acceptable.
Someone /could/ go and ask the developers what their intentions were,
and if they're fine w/ removal for completely non-commercial reasons. It
won't be me since they barely ever responded to me about anything in the
first place. (But others seem to have this problem as well.)
--zeurkous.
--
Friggin' Machines!
