Because it is not.

(In reply to Chris Wilson from comment #1)
> Because it is not.

How can you say this when looking at the reddit thread over there:

https://www.reddit.com/r/archlinux/comments/4cojj9/it_is_probably_time_to_ditch_xf86videointel/

?

There surely must be something to it after reading all those comments?

Because I know of the feature regressions, incompatilibities, lack of suport and bugs in the code.

xf86-video-intel is not stable for me (GPU hangs in Chrome - reported).  Modesetting is rock solid.

Two identical runs of gtkperf, mesa, xorg and all drivers straight out of git:

First modesetting:

GtkPerf 0.40 - Starting testing: Wed Apr  6 08:38:33 2016

GtkEntry - time:  0.01
GtkComboBox - time:  0.94
GtkComboBoxEntry - time:  0.77
GtkSpinButton - time:  0.06
GtkProgressBar - time:  0.16
GtkToggleButton - time:  0.32
GtkCheckButton - time:  0.10
GtkRadioButton - time:  0.17
GtkTextView - Add text - time:  0.21
GtkTextView - Scroll - time:  0.00
GtkDrawingArea - Lines - time:  0.56
GtkDrawingArea - Circles - time:  1.77
GtkDrawingArea - Text - time:  0.12
GtkDrawingArea - Pixbufs - time:  0.04
 --- 
Total time:  5.23

Now xf86-video-intel:

GtkPerf 0.40 - Starting testing: Wed Apr  6 08:41:32 2016

GtkEntry - time:  0.01
GtkComboBox - time:  1.75
GtkComboBoxEntry - time:  1.62
GtkSpinButton - time:  0.07
GtkProgressBar - time:  0.16
GtkToggleButton - time:  0.30
GtkCheckButton - time:  0.10
GtkRadioButton - time:  0.16
GtkTextView - Add text - time:  0.20
GtkTextView - Scroll - time:  0.00
GtkDrawingArea - Lines - time:  0.92
GtkDrawingArea - Circles - time:  1.63
GtkDrawingArea - Text - time:  0.12
GtkDrawingArea - Pixbufs - time:  0.02
 --- 
Total time:  7.05

Should add that is Haswell desktop; Haswell mobile results are very similar.

(In reply to Chris Wilson from comment #3)
> Because I know of the feature regressions, incompatilibities, lack of suport
> and bugs in the code.

Again, at least according to the reddit thread:

https://www.reddit.com/r/archlinux/comments/4cojj9/it_is_probably_time_to_ditch_xf86videointel/

and Phoronix thread:

https://www.phoronix.com/forums/forum/phoronix/latest-phoronix-articles/863332-intel-s-unreleased-3-0-x-org-driver-gets-more-fixes-for-dri3-present

it looks like there are more bugs in xf86-video-intel than in xf86-video-modesetting.

Besides, what I was hinting at was:

(In reply to Chris Wilson from comment #3)
> Because I know of the feature regressions, incompatilibities, lack of suport
> and bugs in the code.

Why not improve on xf86-video-modesetting then instead of having two separate drivers? Just want to understand this.

The KODI and LibreELEC developers apparently are claiming that Intel developers would recommend to use xf86-video-modesetting instead of xf86-video-intel, see:

https://github.com/LibreELEC/LibreELEC.tv/pull/136

And they are also recommending their users to stop using xf86-video-intel, see:

http://forum.kodi.tv/showthread.php?tid=269815&pid=2315595#pid2315595
http://forum.kodi.tv/showthread.php?tid=269815&pid=2315913#pid2315913

Meanwhile, Phoronix did some more xf86-video-intel vs. xf86-video-modesetting benchmarks, see:

http://www.phoronix.com/scan.php?page=news_item&px=Intel-DDX-May-Tests

Regards

And here we go:

https://tjaalton.wordpress.com/2016/07/23/intel-graphics-gen4-and-newer-now-defaults-to-modesetting-driver-on-x/
https://www.phoronix.com/forums/forum/linux-graphics-x-org-drivers/intel-linux/886898-ubuntu-debian-abandon-intel-x-org-driver-for-most-hardware-moves-to-modesetting-ddx

;)