Headless glass platform

[Johan Vos]

Hi Sean,

On Tue, Jan 30, 2024 at 5:09 PM Sean M Phillips <sean.mi.phillips at gmail.com>
wrote:

> This is very cool Johan.  I have some questions if you have time to answer:
>
> Do you believe your headless platform POC would allow the use of the
> snapshot method for exporting visuals of a scene to a png?
>

Yes, that is one of the usecases indeed. I tested it with a print job, and
that worked. So snapshots should work too.


> If so do you have any early indications of performance therein? (number of
> snapshots per second etc)
>

I don't, but keep in mind that the headless platform is only a replacement
in Glass. The rendering part is done in Prism. The headless platform does
not interfere with that. So if the performance of your application is
rendering-bounded, I expect it to be the same. However, since you don't
need a window system with the headless platform, it's easier to run
applications on headless cloud instances, for example.


> Will the headless platform support the 3D Nodes/Camera etc?
>

Yes, that is outside the Glass platform.

>
> Thank you for your work and your time answering in advance.
>
> Sean
>
> On Tue, Jan 30, 2024 at 6:47 AM Johan Vos <johan.vos at gluonhq.com> wrote:
>
>> Hi,
>>
>> I created a branch in the jfx-sandbox repository for experimenting with a
>> headless glass platform:
>> https://github.com/openjdk/jfx-sandbox/tree/johanvos-headless
>>
>> This addresses https://bugs.openjdk.org/browse/JDK-8324941 where I
>> suggest a POC for a Headless platform.
>>
>> There are a number of usecases for this, including:
>> 1. applications that require JavaFX rendering without presenting this to
>> a window (and instead send it to a printer for example)
>> 2. running tests without requiring a window manager.
>>
>> Regarding the second usecase, we already did some basic experiments using
>> a modified version of TestFX where instead of the Monocle Headless
>> subplatform, the POC Headless platform is used.
>>
>> By using a first-class Headless glass platform instead of a Monocle
>> subplatform, it should be easier to use by developers.
>> The monocle code contains very platform/os specific parts, which often
>> don't make sense outside the target platform. This is very valuable, but it
>> is also a very different usecase than a headless platform and it requires a
>> much more complex build procedure.
>>
>> I added an initial, limited HeadlessRobot to do some basic tests. That
>> code is mainly taken from the existing Monocle implementation, but I want
>> to be careful to avoid anything that is not applicable to the headless
>> scenarios.
>>
>> - Johan
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/openjfx-dev/attachments/20240131/d7adf59c/attachment.htm>