Microsoft Visual TrueType Version 6.34 [December, 2020 release]

Paul Linnerud

Download is available here. 

What’s new?

Version 6.34 [December, 2020 release]

The following are changes since 6.33 [November, 2019 release]

• Per display DPI awareness. Support per display DPI when rendering visual elements. This is noticeable when dragging VTT from one display to another on a multi display system.
• Various bug fixes.
  • Fix undo/redo that was not working in the main view.
  • Fix issue with composite detection change detection in variation font.
  • Fix issue where VTT was not opening on display where last closed.
  • Fix IDEF definition limit issue.
  • Fix issue with cvar table generation where tuple records have more than 63 CVT entries.
  • Fix issue with Light Latin auto hinter on complex glyphs.

Viktor Rubenko

Is there any chance that it will be possible to autohint through a shell in future?

Paul Linnerud

We have talked about adding autohinting from VTTShell. Possibly in the future. 

Viktor Rubenko

I started getting bitmap error for size run with random glyphs.

After wich whole interface starts to bug

Piotr Grochowski

What are you expecting from type designers?

Paul Linnerud

@Piotr Grochowski, what version of Windows are you using?

Paul Linnerud

@Piotr Grochowski the features added to VTT in this release require the July 2016 version of Windows 10 or newer (Windows 10, version 1607). 

Piotr Grochowski

Paul Linnerud said:

@Piotr Grochowski the features added to VTT in this release require the July 2016 version of Windows 10 or newer (Windows 10, version 1607). 

I use Windows 7 and I do not trust the VTT update because it is most likely Microsoft telemetry code.

Paul Linnerud

@Piotr Grochowski not sure what you mean by Microsoft telemetry code. VTT does not send any information to Microsoft. 

Paul Linnerud

@Victor Rubenko, I'm unable to reproduce the "Error allocating bitmap for size run" error. 

Piotr Grochowski

Paul Linnerud said:

@Piotr Grochowski not sure what you mean by Microsoft telemetry code. VTT does not send any information to Microsoft. 

If you're so sure about that then write out all of the source code.

Aaron Bell

😂🤣😂🤣

Dave Crossland

Or make it a libre project

Aaron Bell

In the eternal words of the vulture, “oh don’t start that again!”

I mean, unless you want a rerun of the “FontValidator” discussion.

Hin-Tak Leung

What's a 'rerun of the "FontValidator" discussion'? 5 years on, I think that it went well in the sense that we have managed to get it to work on more platforms than before; it did not work well in the sense that it is a rather solitary task...

With regard to the one part that causes the largest amount of fiction - the new freetype backend and the missing original fscaler backend: a few months(?) ago with the news about leaked windows XP code, and some random guy pestering the Openssl people etc to implement font-signing and citing the leaked source (the pestering was/is on github Openssl issue tracker and in the opentype spec issue tracker too), I had a quick look and found not only the original font-signing authentication code, but also fscaler code too, with early ClearType (v37). And it looks some of fontval's adaptation of the win32k.sys fscaler code is in the leaked XP code. So as a private/academic exercise, I can probably recreate the original fscaler backend, and dropping freetype...

All in all, I think it went okay, although there are lessons to be learned and I wish matters done differently if VTT goes libre.

Aaron Bell

@Hin-Tak Leung 

> largest amount of fiction
That's a nice, whitewashed way of referring to the fact that:

1) I was very forward with you (and all involved) that the renderer backend is Microsoft IP and would not be open sourced. Therefore, that the source code is broken. Everyone understood and was OK with that fact. 

2) Despite this, you constantly insisted that Microsoft was 'hiding' a newer version of the source (hint, we weren't) because the source was broken and missing the rasterizer. And then constantly peppered me with emails demanding that I give you the newer source.

3) You stated, both implied and outright, that I was lying to you regarding #2. 

4) Publicly badmouthed Microsoft for the above perceived slights. 

5) Made me so regret pursuing open sourcing Font Validator that I refused to consider any other similar Open Sourcing requests. 

So yes, definitely lessons learnt. 

PS. I would strongly recommend obtaining Microsoft's permission before working with leaked Windows XP code. 

Hin-Tak Leung

There was a newer version of Font Validator within Microsoft than was the last public one. I would have liked to have the last full binary build, for private reference. That had been my wish and still is.

Whatever not available in source form is, just isn't.

FWIW, re-creating the last internal build of FontVal from leaked XP source is just an idea - if I do go ahead with it in a private manner, the result is unlikely to ever see day light, for obvious reasons - it is just interesting to know/see what "would have been".

I am a little curious @Aaron Bell that you feel more strongly about this than I am - I spent 3 years filling in the missing pieces; would have liked a bit of help from many parties (you and Microsoft folks no less) but I have managed okay without.

Hin-Tak Leung

It is perhaps interesting at one point, I felt similar about @Dave Crossland and the Google folks on their pestering about the replacement freetype-based backend, which is a interface-compatible drop-in replacement to the original windows win32k.sys fscaler backend. Well, as I wrote, it is the part that caused/causes the largest amount of fiction.

Aaron Bell

Hin-Tak Leung said:

There was a newer version of Font Validator within Microsoft than was the last public one. I would have liked to have the last full binary build, for private reference. That had been my wish and still is.

As I said then, and say again now, there wasn't. That you still continue to disbelieve me is astounding. 

Hin-Tak Leung said:

I am a little curious @Aaron Bell that you feel more strongly about this than I am - I spent 3 years filling in the missing pieces; would have liked a bit of help from many parties (you and Microsoft folks no less) but I have managed okay without.

I was very (very) clear upfront what I could offer in terms of Microsoft involvement. That was OK with all parties at the start of the project. 

And yes, I do still feel strongly about all this. You publicly maligned me, and the company, with no reason, and never offered an apology for your behavior nor comments. From what you've written above, you still see yourself in the right, and justified. So I have nothing more to say. 

Hin-Tak Leung

@Aaron Bell I believe that you believed what you wrote; and I also believe that you are misinformed/misled, either knowingly or unknowingly, by your colleagues. I don't disbelieve you.

The state of the source code released, though incomplete, is quite a bit ahead (by a few years, 4-6 years by my estimate) compared to the last Microsoft public release. That is a fact.

I am sorry that you feel misaligned. I think you were/are misinformed/misled by your colleagues.

Viktor Rubenko

I face this interface bug over and over again.

https://youtu.be/wpss5e3vQNk

I need to close-open VTT to fix it, but sometimes after closure VTT window doesn't show up like its window height is zero

Remove VTT entry from regestry helps with this.

Also often text selection in ControlProgram doesn't color blue, it just keeps be transparent.

Viktor Rubenko

Variation CVT shows strange set of masters from gvar.

Real masters from fontlab:

VTT:

Thomas Phinney

Hin-Tak Leung said:

The state of the source code released, though incomplete, is quite a bit ahead (by a few years, 4-6 years by my estimate) compared to the last Microsoft public release. That is a fact.

Hin-Tak, do you mean to say behind, rather than ahead?

Paul Linnerud

@Viktor Rubenko can you provide me with font?

Viktor Rubenko

@Paul Linnerud
sent PM

Paul Linnerud

Thanks, there is code in VTT that collects unique tuple coordinates from gvar and presents them in that list in Variation CVT view as potential locations for cvar coordinates. Then named instance coordinates (if available) are normalized and compared to collected tuple coordinates to present better names in the master list if matches are found. There seems to be issues finding the matches either because the tuples in gvar don't corresponds to named instances or something is going wrong with VTTs coordinate normalization or comparison code. I'm still looking into it. 

Hin-Tak Leung

Thomas Phinney said:

Hin-Tak Leung said:

The state of the source code released, though incomplete, is quite a bit ahead (by a few years, 4-6 years by my estimate) compared to the last Microsoft public release. That is a fact.

Hin-Tak, do you mean to say behind, rather than ahead?

I did mean "ahead". The last Microsoft public binary release was dated 2003; the incomplete source code released in 2015, included work-in-progress changes for the 2009 opentype spec, and probably abandoned around 2007. It is newer by 4-6 years.

Paul Linnerud

@Viktor Rubenko, I sent at PM after looking at your test font. 

Viktor Rubenko

Paul Linnerud said:

@Viktor Rubenko, I sent at PM after looking at your test font. 

Thanks for the hard work.

Viktor Rubenko

@Paul Linnerud

CVT Variation doesn't affect on contours if entries with number 256 and higher are edited, is it maximum or bug?

and I faced this during tests

Viktor Rubenko

Also, is it possible not to reset scroll position in the CVT variation window after compilation?
With a large number of entries, you have to scroll down after each compilation.
And enable ctrl+H from the Variation preview window.