[Freeglut-developer] Fwd: Unifying GLES1 and GLES2

[Diederick C. N. <dc...@gm...>]

Hi Guys,

See Matti's response below, accidentally only sent to me. Sylvain, others,
do you have some thoughts on the below points?

Thanks!
Dee

---------- Forwarded message ----------
From: Matti Lehtonen <m-M...@ik...>
Date: Mon, Jul 7, 2014 at 9:06 AM
Subject: Re: [Freeglut-developer] Unifying GLES1 and GLES2
To: "Diederick C. Niehorster" <dc...@gm...>


Hi,


On 3 July 2014 21:24, Diederick C. Niehorster <dc...@gm...> wrote:

> Hi Matti,
>
> Sorry for the quiet, wasn't in town. Where did you send the patch?
>

Sorry about my late response, I were without a network connection for over
a week because of a move.

I did send the patch as an attachment in this thread, but my suggestion
before the move were that it should be ignored and after the move I have
stronger feelings that the patch should be ignored (for a moment).

Reasoning: Thought it can be used to build a  library with ES1 and ES 3
unificated, I cannot verify that the build is correct.  During block out, I
figured out that my tables do not support OpenGL ES 3 through it is a API
18 device :/

I found this lack of support thru very peculiar way. After a week or so of
hard debugging of strange errors, I decided to debug the beginning of
execution of native code, and I did find out that creation of context of ES
3 failed. Unfortunately I, did not have defined an initialization error
function, so there were no log lines telling of this problem. Apparently
the error message were directed to /dev/stderr, which is basically the same
as /dev/null in Android. It would be really nice, if this behaviour is
changed so (platform dependency) that fgError() and fgWarning() messages
are readable by e.g. logcat in Android even without a custom handler.

Another noticed issue: The fgError() exits by calling exit() function call.
Unfortunately my application were multi activity application and the
application were leaved in kind of half-dead state for a moment (I am not
sure, if the application were actually killed by those side effect errors
or is there some delay before all other activities are killed by Android).
But any way, my other activities did have references to static objects in
that dead activity, causing very strange errors, when the rest of
activities were still running and one activity were killed. It would be
again really nice, if the freeglut causes all activites to be killed in
Android (yet another platform dependency) application in case of error.


Matti L
-- 
Life is complex. It has real and imaginary parts.