8.4.14+cvs on Mac OS X

8.4.12 on Ubuntu/feisty

Please take screenshots of fonttest.tcl on 8.5, I'd be interested to see how it compares. Also, if you want to do the measurements that would be also good to have.

.hc

On Sep 22, 2007, at 6:27 AM, federico wrote:

which is your version of Tk? please note that when I switched to Tk-8.5 with xft support (anti-aliased fonts) I found even bigger fonts while looking at my old tcl/tk projects

shall we collect again the information, for filling a matrix of font sizes?

note: the matrix this time would have to be the cartesian product of: (win32 linux macos)x(<=tk8.4 tk8.5xft)x(font sizes...)x(pd version?)

On 9/22/07, Hans-Christoph Steiner hans@eds.org wrote:

...

I decided to take Pd out of the equation, so I did a quick test using just Tcl/Tk on Mac OS X and Ubuntu Studio. It looks like it confirms what I am talking about: for some reason 12 point is big on GNU/ Linux. So I think we can handle this by forcing 12 point fonts to actually use 11 point on GNU/Linux machines.

Here is the script and the screenshots:

http://pow.idmi.poly.edu/~hans/pdfonts/fonttest/

.hc

---

---

---

If you are not part of the solution, you are part of the problem.

---

PD-dev mailing list PD-dev@iem.at http://lists.puredata.info/listinfo/pd-dev

---

PD-dev mailing list PD-dev@iem.at http://lists.puredata.info/listinfo/pd-dev

------------------------------------------------------------------------ ----

Mistrust authority - promote decentralization. - the hacker ethic

You can get it in the link I posted at the start of this thread:

http://pow.idmi.poly.edu/~hans/pdfonts/fonttest/fonttest.tcl

.hc

On Sep 22, 2007, at 6:15 PM, federico wrote:

sorry but I couldn't find fonttest.tcl on my computer, and even with google. can you post a link, or the script itself?

On 9/22/07, Hans-Christoph Steiner hans@eds.org wrote:

...

8.4.14+cvs on Mac OS X

8.4.12 on Ubuntu/feisty

Please take screenshots of fonttest.tcl on 8.5, I'd be interested to see how it compares. Also, if you want to do the measurements that would be also good to have.

.hc

On Sep 22, 2007, at 6:27 AM, federico wrote:

...

which is your version of Tk? please note that when I switched to Tk-8.5 with xft support (anti-aliased fonts) I found even bigger fonts while looking at my old tcl/tk projects

shall we collect again the information, for filling a matrix of font sizes?

note: the matrix this time would have to be the cartesian product of: (win32 linux macos)x(<=tk8.4 tk8.5xft)x(font sizes...)x(pd version?)

On 9/22/07, Hans-Christoph Steiner hans@eds.org wrote:

...

I decided to take Pd out of the equation, so I did a quick test using just Tcl/Tk on Mac OS X and Ubuntu Studio. It looks like it confirms what I am talking about: for some reason 12 point is big on GNU/ Linux. So I think we can handle this by forcing 12 point fonts to actually use 11 point on GNU/Linux machines.

Here is the script and the screenshots:

http://pow.idmi.poly.edu/~hans/pdfonts/fonttest/

.hc

---

--

---

If you are not part of the solution, you are part of the problem.

---

PD-dev mailing list PD-dev@iem.at http://lists.puredata.info/listinfo/pd-dev

---

PD-dev mailing list PD-dev@iem.at http://lists.puredata.info/listinfo/pd-dev

---

---

---

Mistrust authority - promote decentralization. - the hacker ethic

---

PD-dev mailing list PD-dev@iem.at http://lists.puredata.info/listinfo/pd-dev

------------------------------------------------------------------------ ----

¡El pueblo unido jamás será vencido!

Hey,

I just added Windows picks of the fonttest.tcl. If you at those images in sequence, you can see there are only really tiny variations on each platform, EXCEPT 12 point on GNU/Linux. Strange, but that's the way it seems to be.

As for the existing size measuring code, I think it was part of the problem. Plus if we can make it work on all platforms without any dynamic sizing, which is a great potential source of bugs, why not just hard code it?

I made two little movies to illustrate the issue:

http://pow.idmi.poly.edu/~hans/pdfonts/fonttest/normal.mov http://pow.idmi.poly.edu/~hans/pdfonts/fonttest/bold.mov

As you can see in the movie, there is no dynamic measurement needed, the fonts are all very close one tk scaling is set to 1. Except, of course, the pesky 12 point on GNU/Linux.

.hc

On Sep 22, 2007, at 2:21 PM, Miller Puckette wrote:

HI HC,

Pd has code automatically to do that, which is disabled in your font patch for some reason:

        /* best is now the host font index for the desired font  

index i. */ sys_fontlist[i].fi_hostfontsize = atom_getintarg(3 * best + 2, argc, argv); sys_fontlist[i].fi_width = atom_getintarg(3 * best + 3, argc, argv); sys_fontlist[i].fi_height = atom_getintarg(3 * best + 4, argc, argv);

cheers Miller

On Sat, Sep 22, 2007 at 01:08:23AM -0400, Hans-Christoph Steiner wrote:

...

I decided to take Pd out of the equation, so I did a quick test using just Tcl/Tk on Mac OS X and Ubuntu Studio. It looks like it confirms what I am talking about: for some reason 12 point is big on GNU/ Linux. So I think we can handle this by forcing 12 point fonts to actually use 11 point on GNU/Linux machines.

Here is the script and the screenshots:

http://pow.idmi.poly.edu/~hans/pdfonts/fonttest/

.hc

---

---

---

If you are not part of the solution, you are part of the problem.

---

PD-dev mailing list PD-dev@iem.at http://lists.puredata.info/listinfo/pd-dev

------------------------------------------------------------------------ ----

If you are not part of the solution, you are part of the problem.

On Sep 22, 2007, at 10:00 PM, Miller Puckette wrote:

Hi HC,

Two issues I don't know how to deal with yet: first, although I agree the fonts you're using seem to have nearly consistent sizes, they are smaller than the "traditional" sizes so everyone's existing patches will look wrong. For compatibilty's sake Pd should somehow read old patches with the old sizes by default.

On my Debian etch machine, the Pd-extended fonts are slightly larger:

http://pow.idmi.poly.edu/~hans/pdfonts/screenshots/hcs/Debian- etch_new-old_compare.png

As for the other platforms, on Windows, the fonts were always too small, on Mac OS X, they were too big, and on GNU/Linux, things are all over the place because things can be configured so many different ways. For example, on my Debian Etch machine, it doesn't have Courier installed (it's not installed by default), so it's using some other font. I think this is pretty common. So basically, there is no existing standard size.

Second, I gather that lucida isn't guaranteed to be installed in linux or other flavors of unix (BSD, anyone?) In particular, the default install of Fedora/Gnome doesn't seem to have it. So Pd has to be able to do something appropriate when lucida isn't there.

Any flavor of Lucida is guaranteed to be installed only on Windows. I wasn't suggesting to use it as the default font on all platforms, only Windows, as long as it is very close in size to Bitstream Vera Sans Mono. For example, Pd-extended uses Monaco as the default font on Mac OS X which is very close in size to Bitstream Vera Sans Mono.

It's a shame they aren't free since they are very nice fonts. They are included with Java, but you probably don't want to make people install Java in order to get the fonts working properly. :D

These are probably quite possible to solve. In a related skirmish, I have to introduce a new file format in some upcoming release to allow for controlling the width of boxes, which, stupidly, I never left a place for in the current format.

That would be nice to have for sure.

.hc

cheers M

On Sat, Sep 22, 2007 at 08:55:19PM -0400, Hans-Christoph Steiner wrote:

...

Hey,

I just added Windows picks of the fonttest.tcl. If you at those images in sequence, you can see there are only really tiny variations on each platform, EXCEPT 12 point on GNU/Linux. Strange, but that's the way it seems to be.

As for the existing size measuring code, I think it was part of the problem. Plus if we can make it work on all platforms without any dynamic sizing, which is a great potential source of bugs, why not just hard code it?

I made two little movies to illustrate the issue:

http://pow.idmi.poly.edu/~hans/pdfonts/fonttest/normal.mov http://pow.idmi.poly.edu/~hans/pdfonts/fonttest/bold.mov

As you can see in the movie, there is no dynamic measurement needed, the fonts are all very close one tk scaling is set to 1. Except, of course, the pesky 12 point on GNU/Linux.

.hc

On Sep 22, 2007, at 2:21 PM, Miller Puckette wrote:

...

HI HC,

Pd has code automatically to do that, which is disabled in your font patch for some reason:

       /* best is now the host font index for the desired font

index i. */ sys_fontlist[i].fi_hostfontsize = atom_getintarg(3 * best + 2, argc, argv); sys_fontlist[i].fi_width = atom_getintarg(3 * best + 3, argc, argv); sys_fontlist[i].fi_height = atom_getintarg(3 * best + 4, argc, argv);

cheers Miller

On Sat, Sep 22, 2007 at 01:08:23AM -0400, Hans-Christoph Steiner wrote:

...

I decided to take Pd out of the equation, so I did a quick test using just Tcl/Tk on Mac OS X and Ubuntu Studio. It looks like it confirms what I am talking about: for some reason 12 point is big on GNU/ Linux. So I think we can handle this by forcing 12 point fonts to actually use 11 point on GNU/Linux machines.

Here is the script and the screenshots:

http://pow.idmi.poly.edu/~hans/pdfonts/fonttest/

.hc

---

--

---

If you are not part of the solution, you are part of the problem.

---

PD-dev mailing list PD-dev@iem.at http://lists.puredata.info/listinfo/pd-dev

---

---

---

If you are not part of the solution, you are part of the problem.

------------------------------------------------------------------------ ----

¡El pueblo unido jamás será vencido!