Re: [MiNT] fVDI drawing bugs

[Eero Tamminen <oak@helsinkinet.fi>]

Hi,

On keskiviikko 12 syyskuu 2012, Paul Wratt wrote:
> Thanks for this explicit list of issues with FVDI.

>From the MyAES page which link you sent, I noticed also this (from 2011):
* "FVDI can crash on some VDI function operation if perimeter is set"

It would be good to have a minimal test-case for that, if somebody
knows more details on how to trigger it.


> I believe all have been discussed at one time or another, but with a list
> there is real chance of getting them fixed

The FVDI repo contains partial.txt which seems to list missing
functionality:
-------------------
Drivers
 Both bitplane ones
  Only horizontal and vertical lines can use XOR mode.
  No line patterns are supported.
  The mouse shape will wrap around from the right to the left edge.
 Monochrome
  Normal text clipping to the right is not done correctly.
 16 colour
  Text, icons, and patterned fills are black and white only.

Engine
 No text scaling is done.
 Thick lines don't get any circles in the corners.
 Polygon peripheries aren't drawn.
 The only marker that's supported is single pixel.
 v_opnbm doesn't work.
-------------------

I'm not sure how up to date the list is, at least some line patterns ARE
supported.

And it doesn't mention that thin lines are missing all line endings
(including arrows) and the problem with filled circles.

I hadn't mentioned the text clipping issue because it's also in EmuTOS,
for italic text.  And the text scaling I also noticed, I just wasn't sure
whether it's just related to resolution dependent text point sizes.


> I know Helmet has a "fixed v_clswk()" version of FVDI, so as soon as
> he posts a patch, or the source file with the change, then it will get
> applied to repos

Did you mean v_clrwk() ("clear screen", not close workstation)?

I noticed it's currently in engine/unimpl.s, and just returns without
doing anything.


> note that there is also a vdi test program in FVDI repo, or on Johans
> FVDI FTP page (Misc)

It's a benchmark, not a correctness tester.

It also seems to bomb if FVDI isn't installed which makes it unusable
for correctness testing as one cannot compare it against normal
TOS.  It bombed also with older (v2.5) NVDI.


	- Eero

> Paul
> 
> On Tue, Sep 11, 2012 at 8:42 AM, Eero Tamminen <oak@helsinkinet.fi> wrote:
> > Hi,
> > 
> > I ran the VDI testers I wrote for EmuTOS and noticed several
> > additional things not working correctly with fVDI (in Aranym
> > AFROS default 8-bit mode):
> > 
> > * VDI markers
> > 
> >   - don't show up at all (or only as dots)
> > 
> > * VDI color query
> > 
> >   - seems to return wrong RGB values
> > 
> > * VDI lines
> > 
> >   - zero width lines don't have line endings (arrows etc),
> >   - wider lines don't implement all line endings (round ones)
> >   - arcs have gaps in them
> > 
> > * VDI circles
> > 
> >   - Fill color is not always correct.
> > 
> > All these work in EmuTOS (in 1-4 bit modes supported by it).
> > 
> > See the attachments for couple of examples.
> > 
> > 
> > For completeness sake here are also the issued I noticed / mailed
> > about earlier:
> > 
> > * v_exit_cur() bus errors (it could be no-op)
> > 
> > * Line drawing in XOR mode doesn't always invert the drawing
> > 
> >   (I have a suspicion that this could be Aranym VDI NatFEATS issue)
> > 
> > * v_clswk() doesn't do anything
> > 
> > 
> > I'll put the the source code and binaries up later this week
> > after I've finished some other things.
> > 
> > If somebody has non-NatFEATS version of latest fVDI, it would
> > be nice to either get it or hear whether some of above ones are
> > actually bugs in Aranym, not fVDI.
> > 
> >         - Eero