Re: [OpenJDK 2D-Dev] Fw: RFR: JDK-8187100: support VariationSelectors(Resend)

2018-06-25 Thread Steven R. Loomis 
Pushed… thanks!

On Mon, Jun 25, 2018 at 11:40 AM Phil Race  wrote:

> Ok. Approved.
>
> -phil.
>
> On 06/25/2018 10:20 AM, Toshio 5 Nakamura wrote:
>
> Hi Phil,
>
> In other words, I just misunderstood your suggestion below.
> Recent discussions led me correct understanding, and I took this way.
> sorry for your confusion.
>
> > For example we could explore having isComplexCharCode return true for
> these and
> > directing all uses down the layout path ? Then we can maybe remove all
> this logic from
> > case used by the fast path code.
>
> Toshio Nakamura
> - Forwarded by Toshio 5 Nakamura/Japan/IBM on 2018/06/26 02:09 -
>
> From: "Toshio 5 Nakamura"  
> To: Phil Race  
> Cc: 2d-dev <2d-dev@openjdk.java.net> <2d-dev@openjdk.java.net>, "Steven
> R. Loomis"  
> Date: 2018/06/23 09:08
> Subject: Re: [OpenJDK 2D-Dev] RFR: JDK-8187100: support Variation
> Selectors(Resend)
> Sent by: "2d-dev" <2d-dev-boun...@openjdk.java.net>
> <2d-dev-boun...@openjdk.java.net>
> --
>
>
>
> Hi Phil, Steven,
>
> Thank you so much for your support, and sorry for taking your time.
>
> As Akira-san suggested, Harfbuzz has VS capability.
> It seems difficult to use its glyph picking up mechanism in this case,
> but its layout mechanism we can use. And, I tried to use it.
>
> Regarding default UVS table, if composite fonts keep font slot,
> we just need to check whether Base+VS is defined in Non-default UVS table
> or not.
> When Non-default UVS doesn't have the entry, we always use its default
> glyph.
>
> Best regards,
>
> Toshio Nakamura
>
> [image: Inactive hide details for Phil Race ---2018/06/23 07:34:06---Oh ..
> so this is doing what I suggested a couple of emails back. I]Phil Race
> ---2018/06/23 07:34:06---Oh .. so this is doing what I suggested a couple
> of emails back. If we see a VS we always delegate t
>
> From: Phil Race  
> To: "Steven R. Loomis"  
> Cc: Toshio 5 Nakamura  , 2d-dev
> <2d-dev@openjdk.java.net> <2d-dev@openjdk.java.net>
> Date: 2018/06/23 07:34
> Subject: Re: [OpenJDK 2D-Dev] RFR: JDK-8187100: support Variation
> Selectors(Resend)
> --
>
>
>
> Oh .. so this is doing what I suggested a couple of emails back.
> If we see a VS we always delegate to TextLayout  ?
> Yes, it does simplify the patch a lot and I am perfectly OK with
> this approach but I am surprised as, but Toshio was very keen on
> having it there without TextLayout being needed .. why the change of heart
> ?
>
> But I don't get the removal (skipping) of the Default UVS table ?
>
> -phil.
>
> On 06/22/2018 03:26 PM, Steven R. Loomis wrote:
>
>Phil, does this help:
>
>
> 
> *https://gist.github.com/srl295/a81ec3a8495d53b85f368a7872138e86#file-webrev-02-vs-03-diff*
> 
> 
>
> it is a diff between the 02 and 03 versions.
>
> On Fri, Jun 22, 2018 at 2:59 PM Phil Race <
> *philip.r...@oracle.com* > wrote:
>Please save me time and tell me where I will find the
>   changes from the .02 version ?
>   In particular I mean where are the changes associated
>   with
>   "Use TextLayout (Harfbuzz) if VS appears." ?
>
>   -phil.
>
>   On 06/22/2018 12:59 PM, Steven R. Loomis wrote:
>  Updated webrev:
>
> *http://cr.openjdk.java.net/~srl/8187100/webrev.03*
>   
> 
>
>   - Use TextLayout (Harfbuzz) if VS appears.
>   - Composite font's behavior was changed to
>   not change the font by VS.
>   These changes made the patch so simplified.
>   - add comment about suggested DejaVu version
>
>
>
>   On Thu, May 31, 2018 at 3:19 PM Steven R.
>   Loomis <*s...@icu-project.org*
>   > wrote:
>   Updated webrev:
>
> *http://cr.openjdk.java.net/~srl/8187100/webrev.01/*
>   
> 
>
>   On Fri, May 18, 2018 at 9:16 AM, Toshio 5
>   Nakamura <*toshi...@jp.ibm.com*
>   > wrote:
>  Thank you for your review, Phil.
> I'm working to handle your points.
>
> Toshio Nakamura
>
> Philip Race <*philip.r...@oracle.com*
> > wrote on
> 2018/05/18 13:39:35:
>
> > From: Philip Race <
>

Re: [OpenJDK 2D-Dev] Fw: RFR: JDK-8187100: support VariationSelectors(Resend)

2018-06-25 Thread Phil Race 

Ok. Approved.

-phil.

On 06/25/2018 10:20 AM, Toshio 5 Nakamura wrote:


Hi Phil,

In other words, I just misunderstood your suggestion below.
Recent discussions led me correct understanding, and I took this way.
sorry for your confusion.

> For example we could explore having isComplexCharCode return true 
for these and
> directing all uses down the layout path ? Then we can maybe remove 
all this logic from

> case used by the fast path code.

Toshio Nakamura
- Forwarded by Toshio 5 Nakamura/Japan/IBM on 2018/06/26 02:09 -

From: "Toshio 5 Nakamura" 
To: Phil Race 
Cc: 2d-dev <2d-dev@openjdk.java.net>, "Steven R. Loomis" 


Date: 2018/06/23 09:08
Subject: Re: [OpenJDK 2D-Dev] RFR: JDK-8187100: support Variation 
Selectors(Resend)

Sent by: "2d-dev" <2d-dev-boun...@openjdk.java.net>





Hi Phil, Steven,

Thank you so much for your support, and sorry for taking your time.

As Akira-san suggested, Harfbuzz has VS capability.
It seems difficult to use its glyph picking up mechanism in this case,
but its layout mechanism we can use. And, I tried to use it.

Regarding default UVS table, if composite fonts keep font slot,
we just need to check whether Base+VS is defined in Non-default UVS 
table or not.
When Non-default UVS doesn't have the entry, we always use its default 
glyph.


Best regards,

Toshio Nakamura

Inactive hide details for Phil Race ---2018/06/23 07:34:06---Oh .. so 
this is doing what I suggested a couple of emails back. IPhil Race 
---2018/06/23 07:34:06---Oh .. so this is doing what I suggested a 
couple of emails back. If we see a VS we always delegate t


From: Phil Race 
To: "Steven R. Loomis" 
Cc: Toshio 5 Nakamura , 2d-dev 
<2d-dev@openjdk.java.net>

Date: 2018/06/23 07:34
Subject: Re: [OpenJDK 2D-Dev] RFR: JDK-8187100: support Variation 
Selectors(Resend)





Oh .. so this is doing what I suggested a couple of emails back.
If we see a VS we always delegate to TextLayout  ?
Yes, it does simplify the patch a lot and I am perfectly OK with
this approach but I am surprised as, but Toshio was very keen on
having it there without TextLayout being needed .. why the change of 
heart ?


But I don't get the removal (skipping) of the Default UVS table ?

-phil.

On 06/22/2018 03:26 PM, Steven R. Loomis wrote:

Phil, does this help:


_https://gist.github.com/srl295/a81ec3a8495d53b85f368a7872138e86#file-webrev-02-vs-03-diff_

it is a diff between the 02 and 03 versions.

On Fri, Jun 22, 2018 at 2:59 PM Phil Race
<_philip.race@oracle.com_
> wrote:
Please save me time and tell me where I will
find the changes from the .02 version ?
In particular I mean where are the changes
associated with
"Use TextLayout (Harfbuzz) if VS appears." ?

-phil.

On 06/22/2018 12:59 PM, Steven R. Loomis wrote:
Updated webrev:
_

__http://cr.openjdk.java.net/~srl/8187100/webrev.03_



- Use TextLayout (Harfbuzz) if
VS appears.
- Composite font's behavior
was changed to not change the
font by VS.
These changes made the patch
so simplified.
- add comment about suggested
DejaVu version



On Thu, May 31, 2018 at 3:19
PM Steven R. Loomis
<_srl@icu-project.org_
>
wrote:
Updated webrev:
_

__http://cr.openjdk.java.net/~srl/8187100/webrev.01/_



On Fri, May 18, 2018 at 9:16
AM, Toshio 5 Nakamura
<_toshi...@jp.ibm.com_
>
wrote:

[OpenJDK 2D-Dev] Fw: RFR: JDK-8187100: support VariationSelectors(Resend)

2018-06-25 Thread Toshio 5 Nakamura 


Hi Phil,

In other words, I just misunderstood your suggestion below.
Recent discussions led me correct understanding, and I took this way.
sorry for your confusion.

> For example we could explore having isComplexCharCode return true for
these and
> directing all uses down the layout path ? Then we can maybe remove all
this logic from
> case used by the fast path code.

Toshio Nakamura
- Forwarded by Toshio 5 Nakamura/Japan/IBM on 2018/06/26 02:09 -

From:   "Toshio 5 Nakamura" 
To: Phil Race 
Cc: 2d-dev <2d-dev@openjdk.java.net>, "Steven R. Loomis"

Date:   2018/06/23 09:08
Subject:Re: [OpenJDK 2D-Dev] RFR: JDK-8187100: support Variation
Selectors(Resend)
Sent by:"2d-dev" <2d-dev-boun...@openjdk.java.net>



Hi Phil, Steven,

Thank you so much for your support, and sorry for taking your time.

As Akira-san suggested, Harfbuzz has VS capability.
It seems difficult to use its glyph picking up mechanism in this case,
but its layout mechanism we can use. And, I tried to use it.

Regarding default UVS table, if composite fonts keep font slot,
we just need to check whether Base+VS is defined in Non-default UVS table
or not.
When Non-default UVS doesn't have the entry, we always use its default
glyph.

Best regards,

Toshio Nakamura

Inactive hide details for Phil Race ---2018/06/23 07:34:06---Oh .. so this
is doing what I suggested a couple of emails back. IPhil Race ---2018/06/23
07:34:06---Oh .. so this is doing what I suggested a couple of emails back.
If we see a VS we always delegate t

From: Phil Race 
To: "Steven R. Loomis" 
Cc: Toshio 5 Nakamura , 2d-dev
<2d-dev@openjdk.java.net>
Date: 2018/06/23 07:34
Subject: Re: [OpenJDK 2D-Dev] RFR: JDK-8187100: support Variation Selectors
(Resend)



Oh .. so this is doing what I suggested a couple of emails back.
If we see a VS we always delegate to TextLayout  ?
Yes, it does simplify the patch a lot and I am perfectly OK with
this approach but I am surprised as, but Toshio was very keen on
having it there without TextLayout being needed .. why the change of
heart ?

But I don't get the removal (skipping) of the Default UVS table ?

-phil.

On 06/22/2018 03:26 PM, Steven R. Loomis wrote:
Phil, does this help:



https://gist.github.com/srl295/a81ec3a8495d53b85f368a7872138e86#file-webrev-02-vs-03-diff


it is a diff between the 02 and 03 versions.

On Fri, Jun 22, 2018 at 2:59 PM Phil Race <
philip.r...@oracle.com> wrote:
  Please save me time and tell me where I will find the
  changes from the .02 version ?
  In particular I mean where are the changes associated
  with
  "Use TextLayout (Harfbuzz) if VS appears." ?

  -phil.

  On 06/22/2018 12:59 PM, Steven R. Loomis wrote:
  Updated webrev:

  http://cr.openjdk.java.net/~srl/8187100/webrev.03


  - Use TextLayout (Harfbuzz) if VS appears.
  - Composite font's behavior was changed to
  not change the font by VS.
  These changes made the patch so simplified.
  - add comment about suggested DejaVu version



  On Thu, May 31, 2018 at 3:19 PM Steven R.
  Loomis  wrote:
  Updated webrev:

  http://cr.openjdk.java.net/~srl/8187100/webrev.01/


  On Fri, May 18, 2018 at 9:16 AM, Toshio 5
  Nakamura  wrote:
Thank you for your review, Phil.
I'm working to handle your points.

Toshio Nakamura

Philip Race 
wrote on 2018/05/18 13:39:35:

> From: Philip Race <
philip.r...@oracle.com>
> To: Toshio 5 Nakamura <
toshi...@jp.ibm.com>
> Cc: 2d-dev <2d-dev@openjdk.java.net>
> Date: 2018/05/18 13:39



> Subject: Re: [OpenJDK 2D-Dev] RFR:
JDK-8187100: support Variation
> Selectors(Resend)
>
> There's a lot to think about here but
I have some requests to first
> clean up the proposed code to conform
to coding standards.
>