[lug] Local Open Source Project Leads/Managers

Wed Mar 14 15:53:20 MST 2001

"Riggs, Robert" wrote:
> 
> Two things:
> 
> Regarding the font problem, generally it occurs when web designers/tools set
> pixel (px) font sizes in their style sheets, instead of point (pt) sizes.
> Your 8 pixel font might look fine on your 1024x768 screen (or under Windows,
> which doesn't really allow pixel sized fonts, per se), but it will look like
> a small smudge on my 1600x1200 screen. One other problem is that Netscape's

FYI, I am running 1600x1200 on one monitor, 1280x1024 on another. I
suspect that Netscape has a flaw when trying to override font sizes and
they are specified in pixel size, as you mention above.

> default font scaling is too severe. Adding the following to my ~/.Xdefaults
> file really helped: "Netscape.documentFonts.sizeIncrement:  15".
> 
> I would recommend that anyone considering CORBAfying their software stay
> away from ORBit, if at all possible. In its current incarnation, it is not a
> fully functioning ORB. However, it is one of the only ORBs with C bindings
> available. I've been using omniORB for a bit and am very happy with it. It
> will work well for anyone needing C++ or Python bindings. I was really glad
> to see this project using it.

It would be interesting to see dual support for both ORBit and omniORB.
My reason being that many gnome/gtk apps will support ORBit without
modification, whereas each app would require optimization to run under
omniORB. What would be even more interesting is if gnome could be made
to run correctly (without quirks) under omniORB, or even to write an
interface to allow omniORB and ORBit to transparently talk to each
other.

> 
> -----Original Message-----
> From: D. Stimits [mailto:stimits at idcomm.com]
> Sent: Wednesday, March 14, 2001 1:01 PM
> To: lug at lug.boulder.co.us
> Subject: Re: [lug] Local Open Source Project Leads/Managers
> 
> Steve Houston wrote:
> >
> > Sorry I have taken so long to reply....
> >
> > > "D. Stimits" <stimits at idcomm.com> wrote:
> > >
> > >
> > > I have an interest in a game project, which is currently under
> > > discussion of redesign. I recently converted old C code to a C++
> > > version, with the intent of redesigning it in parts, which is turning
> > > out to be slow. But the short explanation is that there is interest in
> > > creating an OpenGL version and starting with a client and server that
> > > use some of the old ideas, but not much of the original code.
> > > http://www.battlefieldlinux.com
> >
> > Cool... that's an interesting project.
> 
> It still has some work to be done before it gains popularity.
> 
> >
> > > Unfortunately, the web page is difficult to read; my Netscape shows the
> > > fonts as tiny and not very good resolution, even though I've tried to
> > > force both fixed and variable width fonts to 24 point (I'd estimate that
> > > http://www.3dwm.org/frameset.html is showing up at approximately the
> > > same size as a 6 point font, and lots of aliasing). Basically, some of
> > > the web pages are not readable (I'd have to do a copy and paste to vi or
> > > something to see what it says). Contrary to all the above whining, I am
> > > able to read the php pages, which apparently use a different font (it's
> > > still small, but I can read most of it).
> >
> > I've let our webmaster know about the problems you had and he'll try and
> > address them shortly.
> 
> I have seen other web sites like this, the webmasters never could figure
> out why it was doing it. Presumably it has something to do with
> Netscape's font handling (or lack there of).
> 
> >
> > > What I am curious about is that this says it is not a window manager.
> > > But it appears it requires an X11 server in the background.
> >
> > Currently X11 is required for 2 reasons:
> >
> > 1) OpenGL acceleration. Essentially, to get good OpenGL driver support
> > you have to be running XFree86 4.0 with DRI. We plan to move away X
> > (mainly because of it's large memory/resource requirements and our
> > interest in wearable computers), which will probably involve SDL or GGI.
> > Currently though, these don't offer the same level of hardware
> > acceleration as X/DRI.
> >
> > 2) Support for regular 2D applications. This is achieved by running X11
> > (either locally or remotely) and displaying the X desktop in our 3D
> > environment using VNC. You can an example of this on the screenshots
> > page.
> >
> > > Is it
> > > possible to run all the regular applications in this environment?
> >
> > Yep :)
> >
> > > Can
> > > something like the gnome ORB and security methods still be used?
> >
> > We aren't using ORBit (which is what gnome uses), instead we use
> > omniORB. We haven't looked at security too much so far, we are still in
> > the initial stages of building the UI. Can you give an example of what
> > you are proposing?
> 
> Basically this was combining two related questions, and is a bit
> misleading the way I asked it. I ask about the gnome ORB because there
> are a lot of applications that use this as a substitute to the COM/DCOM
> model in windows. Anything working correctly with the gnome ORB has an
> advantage for inter-application communications. In terms of security,
> xauth and xhost mechanisms (optionally kerberos) can be used with X11,
> and I think the gnome ORB might have some knowledge of this to allow it
> to work smoothly with those methods.
> 
> >
> > > Requirements show the need for cavelib to be installed, but not
> > > multiprocessor...I happen to run a couple of SMP machines, I'm curious
> > > if the 3dwm works well with SMP?
> >
> > Hmmm, that shouldn't be in the requirements. Where did you see that, I'd
> > like to fix it. You only need cavelib if you are actually running in a
> > CAVE.
> 
> Apparently I saw this in the 3D-CUBE listing, and assumed it went with
> the rest of the items, but in reality does not.
> 
> >
> > We haven't done any testing on an SMP system, however I believe it will
> > beneficial. 3dwm uses a threaded client/server architecture. The server
> > is responsible for rendering the 3D environment and the client(s) are
> > responsible for generating the world content (building the scene graph).
> > I think this should scale nicely on multiple CPUs.
> >
> > > I guess my main problem with trying this all out is that cavelib appears
> > > to be commercial only.
> >
> > Our mistake, it will run fine without cavelib.
> >
> > Regards,
> > Steve.
> 
> I'm curious if the future version, that does not require X11 but runs on
> Linux, will have support for multiple or remote displays?
> 
> Also, has any kind of benchmarking been done? I'm curious if this
> dedicated 3D has any speed advantages over hardware OpenGL on a regular
> X11 display? It seems like the possibility is there, I like the idea a
> lot.
> 
> D. Stimits, stimits at idcomm.com
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug