The absolute position stuff is 100%. It was only a few lines of code. (One
clarification though: it is in the server's logical coordinates that, now that
rotation is added, may or may not be physical coordinates. Is this a probelm for the
touchscreen calibration?)
Screen rotation works except for vertical text (which when rotated is actually
upside-down) and scrolling. I finished up most of the blits this afternoon on my
laptop and after running a couple tests I'll commit it.
Rotating the screen at runtime is very similar to changing the depth at runtime, I'll
work on that after I finish the theme fix and rotation itself.
As I type I'm taking my second stab this week at fixing the theme bug. I was a bit
worried when I was trying to fix it at school during my independant study, and it
would randomly crash. I have confirmed that this was a hardware problem, and with a
better power supply it works find.
I'm pretty convinced I can fix it this week, hopefully today or tomorrow. I just ran a
lengthy trace of the theme fillstyle interpreter, and this section particularly caught
my eye:
FILLSTYLE --- Op: 0x10 Stk: 0 0 65 13 0
FILLSTYLE --- Op: 0x11 Stk: 0 0 65 13 0 0
FILLSTYLE --- Op: 0x12 Stk: 0 0 65 13 0 0 65
FILLSTYLE --- Op: 0x13 Stk: 0 0 65 13 0 0 65 13
Local theme lookup, property 1
theme_lookup(0x0003,0x0001) = 0x00C0C0C0
FILLSTYLE --- Op: 0x2D Stk: 0 0 65 13 0 0 65 13 12632256
FILLSTYLE --- Op: 0x00 Stk: 12632256 0 65 13 0 0 65 13
FILLSTYLE --- Op: 0x01 Stk: 12632256 13 65 13 0 0 65
FILLSTYLE --- Op: 0x2E Stk: 12632256 13 65 13 0 0 1
+36 #191 (41448) 0x0070D9A0
The first four lines are simply making a copy of the original coordinates. The theme
makes a lookup, which seems to be sucessful. That long ugly number is a RGB color in
decimal. It then proceeds to execute an extra opcode that sets the original X
coordinate to this color!
This would cause all gropnodes to either die with a stack underflow or be placed off
the edge of the screen, hence the current uneventful output of the theme interpreter.
This extra opcode is likely an off-by-one error in the code I wrote to get around the
m68k's word alignment issues.
On Tue, 20 March 2001, Pascal Bauermeister wrote:
>
> We seem to be deep in the phase nr 1
>(http://cvs.sourceforge.net/cgi-bin/cvsweb.cgi/~checkout~/pgserver/log/pascal-todo.txt?content-type=text/plain&cvsroot=pgui).
>
> I wanted some feedback on my suggestion about resource allocation; to me, phase 1
>was consisting essentially of answers to questions. Instead, there were some points
>solved, some other half solved, some left w/o answer...
>
> Well... Alright. But I need to know exactly the status of the project, and precise
>answers to the questions below.
>
> _______________________________________________
>
> A01: Theme refresh bug
> Micah, can you correct it this week or shall we go with this bug in the pre-release
>? (if unsure, say no. I need to know)
>
> A02: Screen rotation
> Seems to be half finished. Blits not yet working.
> Micah:
> - what about the client request to do the switching ?
> - Do the input drivers also need to be notified, or does
> the server swap x/y ?
> - What will be done (for sure) for the pre-release ?
>
> A03: Depth switching
> Postponed, unless other A points completely finished.
>
> A04: Getting abs position
> 100% finished and tested. Micah, can you confirm this ?
>
> _______________________________________________
>
>
> I'd like to do the pre-release by end of this week, and the release by end of this
>month.
>
> Let's focus on the pre-release.
>
> This pre-release is only to have a global well-known state of the project at a given
>date, so that we can tag the whole project. From then, we'll be able to continue the
>works, but also roll-back temporarily several parts of the project if necessary.
>Never mind if there are still old known bugs (better than new unknown ones), as far
>as the system is predictable. This is vital for us since our marketing guys will go
>on tour with a demo relatively soon, so I'd prefer to do multiple small steps until
>then.
>
> Blocking points are A02 and especially A01. Well, since A02 is on its way, it should
>reach a stable state for the pre-release.
>
> Last question: Micah, do you have an up-to-date list of annoying other bugs or
>missing things I should be aware of ?
>
> Okay, things go just smoothly, actually. I just prefer no hurry but a good
>communication between all of us.
>
> Last note: I recently discovered Micah's log (that's cool), and from now on, I'll be
>logging my checkins in my one too ('pgserver/log/pascal-log.txt' which has a link on
>the home page http://pgui.sourceforge.net/). I'm using the changelog format to
>summarize not only the actions done, but also the files involved. Micah, could you
>please create a similar link for Philippe ? thanks.
>
> Thank you everybody for the good work,
>
> Pascal
>
>
> --
> Pascal Bauermeister
> Head of Software Development
>
> SMARTDATA
> PSE-A / EPFL
> CH-1015 Lausanne
>
> http://www.smartdata.ch
> mailto:[EMAIL PROTECTED]
> Phone: +41 (21) 693 84 98
> fax: +41 (21) 693 84 91
>
> _______________________________________________
> Pgui-devel mailing list
> [EMAIL PROTECTED]
> http://lists.sourceforge.net/lists/listinfo/pgui-devel
--
Only wimps use tape backup: _real_ men just upload
their important stuff on ftp, and let the rest of the
world mirror it ;)
-- Linus Torvalds
_______________________________________________
Pgui-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/pgui-devel
