[MPEG-OTSPEC] lookup processing and 'chws', 'vchw'

[Peter Constable]

NP. My underlying concern is that the spec should not contradict itself or be ambiguous in ways that undermine interoperability.


Peter

-----Original Message-----
From: Ken Lunde <ken.lunde at gmail.com> 
Sent: Saturday, August 22, 2020 5:58 AM
To: Peter Constable <pgcon6 at msn.com>
Cc: MPEG OT Spec list (mpeg-otspec at lists.aau.at) <mpeg-otspec at lists.aau.at>
Subject: Re: [MPEG-OTSPEC] lookup processing and 'chws', 'vchw'

Thank you, Peter, for doing the deep dive into the history and details of this particular wording. I look forward to reviewing the proposed revised text.

Regards...

-- Ken

> On Aug 21, 2020, at 11:48, Peter Constable <pgcon6 at msn.com> wrote:
> 
> I was able to confirm that this wording was present in the OT 1.25 timeframe. The OT spec change log captures that there were “clarifications” made in the description of ‘kern’ in OT 1.01, 1.2 and 1.25, so it seems most likely the wording was introduced in one of those. I haven’t yet tracked those down.
>  
> But in the meantime, I got info offline from Apple and from Microsoft making clear that their implementations don’t do anything feature-specific in terms of how lookups get processed. I’ve considered that together with the info Ken Lunde providing (that he adapted descriptions for ‘chws’ and ‘vchw’ from existing descriptions for ‘kern’ and ‘vkrn’), as well as the input from John Hudson on my related question (how ‘kern’ type 2 lookups typically get implemented in fonts
>  
> That combined feedback is leading me to conclude that what was intended was actually not something for application / layout engine implementers to do as currently suggested by the text in question being in the “Application interface” portion of the feature descriptions. Rather, it appears what was likely intended was guidance to font implementations: when crafting type 2 lookups for ‘kern’/’vkrn’/’chws’/’vchw’, make sure to have the valueFormat2 field set to 0 so that the second glyph is not “consumed” as the lookup is processed over the glyph sequence but becomes the “next” glyph (first glyph for the next pair in the sequence. Otherwise, only every other pair of glyphs can be kerned.
>  
> With that in mind, I’ll draft proposed text for revised feature descriptions for those four features.
>  
>  
> Peter
>  
> From: Adam Twardoch <adam at twardoch.com> 
> Sent: Thursday, August 20, 2020 3:18 PM
> To: Dave Crossland <dcrossland at google.com>
> Cc: MPEG OT Spec list (mpeg-otspec at lists.aau.at) <mpeg-otspec at lists.aau.at>; Peter Constable <pgcon6 at msn.com>
> Subject: Re: [MPEG-OTSPEC] lookup processing and 'chws', 'vchw'
>  
> OpenType 1.25 was last to support MM, and it was dropped in OT 1.3. 
>  
> The entire OT 1.25 spec is available at: 
> https://web.archive.org/web/20000914162807/http://www.microsoft.com/typography/otspec/otsp125.zip 
>  
> A.
>  
> On Thu, 20 Aug 2020 at 23:54, Dave Crossland <dcrossland at google.com> wrote:
>  
> 
> On Thu, Aug 20, 2020, 12:26 AM Peter Constable <pgcon6 at msn.com> wrote:
> 
> I no longer have access to any OT versions prior to OT 1.4. 
>  
> Adam mentioned recently that he had accessed earlier versions of the ot spec with the mm stuff that was dropped by 1.4.
>  
> I seem to recall David Lemon said many years ago there were other things in 1.0-1.2 era that were dropped. 
>  
> I would love to see these old versions be available for historical/academic purposes :)
>  
> Adam, care to reveal your secrets for the plucky historians? :)
>  
> 
>  
> 
> -- 
> Adam Twardoch
> http://www.twardoch.com/
> _______________________________________________
> mpeg-otspec mailing list
> mpeg-otspec at lists.aau.at
> https://lists.aau.at/mailman/listinfo/mpeg-otspec