xinput: GetDeviceButtonMapping-reply: add align-pad at end

see:
http://cgit.freedesktop.org/xorg/lib/libXi/tree/specs/encoding.xml#n1691
(file-position may become wrong if the spec changes)

xinput: GetDeviceKeyMapping-request: add pad at end

see:
http://cgit.freedesktop.org/xorg/lib/libXi/tree/specs/encoding.xml#n1583
http://cgit.freedesktop.org/xorg/proto/inputproto/tree/XIproto.h#n982
(file-positions may become wrong if the spec or header change)

xinput: SetDeviceFocus-request: add pad at end

see:
http://cgit.freedesktop.org/xorg/lib/libXi/tree/specs/encoding.xml#n1308
http://cgit.freedesktop.org/xorg/proto/inputproto/tree/XIproto.h#n737
(file-positions may become wrong if the spec or header change)

xinput: AllowDeviceEvents-request: add pad at end

see:
http://cgit.freedesktop.org/xorg/lib/libXi/tree/specs/encoding.xml#n1251
http://cgit.freedesktop.org/xorg/proto/inputproto/tree/XIproto.h#n692
(file-positions may become wrong if the spec or header change)

xinput: UngrabDeviceButton-request: add pad at end

see:
http://cgit.freedesktop.org/xorg/lib/libXi/tree/specs/encoding.xml#n1229
http://cgit.freedesktop.org/xorg/proto/inputproto/tree/XIproto.h#n676
(file-positions may become wrong if the spec or header change)

xinput: UngrabDevice-request: add pad at end

see:
http://cgit.freedesktop.org/xorg/lib/libXi/tree/specs/encoding.xml#n1130
http://cgit.freedesktop.org/xorg/proto/inputproto/tree/XIproto.h#n596
(file-positions may become wrong if the spec or header change)

xinput: OpenDevice-reply: add align-pad at end

see:
http://cgit.freedesktop.org/xorg/lib/libXi/tree/specs/encoding.xml#n834
(file-position may become wrong if the spec changes)

xinput: SetDeviceModifierMapping: fix length of pad

fix length of pad from 1 to 2 bytes after request-field keycodes_per_modifier

see XIproto.h:1057 in inputproto-2.3.1
see spec in libXi-1.7.4 ( in docbook format in the directory "specs" )

Reviewed-by: Peter Harris <pharris@opentext.com>
Reviewed-by: Ran Benita <ran234@gmail.com>

xinput: ChangeFeedbackControl: add missing pad

add missing pad after field "feedback_id"

see XIproto.h:875 in inputproto-2.3.1
see spec in libXi-1.7.4 ( in docbook format in the directory "specs" )

Reviewed-by: Peter Harris <pharris@opentext.com>
Reviewed-by: Ran Benita <ran234@gmail.com>

Release xcb-proto 1.11

xkb: Add missing modLatches as comment to LatchLockState

The field modLatches was missing in the request LatchLockState:
    http://cgit.freedesktop.org/xorg/proto/kbproto/tree/XKBproto.h#n141

v2: Use a pad instead of a field, as the field would cause an API break.
    Though, keep the field commented out to document it.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com>

xinput: Add missing pad in XIChangeHierarchy

A pad of 3 bytes was missing in the request XIChangeHierarchy:
    http://cgit.freedesktop.org/xorg/proto/inputproto/tree/XI2proto.h#n439

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Peter Harris <pharris@opentext.com>

xinput: Add missing pad in DeviceResolutionCtl

A pad of 2 bytes was missing in the struct DeviceResolutionCtl:
    http://cgit.freedesktop.org/xorg/proto/inputproto/tree/XIproto.h#n1382

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Peter Harris <pharris@opentext.com>

Add <fd> to xml-xcb.txt

Signed-off-by: Ran Benita <ran234@gmail.com>
Reviewed-by: Daniel Martin <consume.noise@gmail.com>

schema: Restricted enum item values

Restrict decimal values by using an xsd:unsignedInt instead of an
xsd:integer. xsd:unsignedInt is an unsigned 32-bit integer.
And restrict bit values by adding a type, which allows values within the
range of [0, 32) only.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

schema: Remove dec-or-hex-integer and hex-integer type

Reflect in the schema that hexadecimal values aren't supported in enum
items (and expressions) by removing the 'dec-or-hex-integer' and
'hex-integer' type.

Hexadecimal values are not supported "as some languages have a different
notation for hex-values" and therefor such hexadecimal values have been
explicitly replaced with decimal values in other commits, see:

    183ecff Replaced hex-values with decimal ones
    8b3c120 did remaining replacements of hex constants with decimal

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

schema: Add attribute 'name' to <fd>

Do not allow any other attribute and make sure the attribute 'name' is
set in an fd element by adding and declaring it as required in the
schema.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

schema: Set default attribute values

Set the default values for the attributes combine-adjacent,
no-sequence-number and xge to false.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

xproto: Fix EnterNotify documentation

Fix a copy'n'paste mistake (from UnmapNotify) in the documentation of
EnterNotify by changing the description for 'event', renaming
'window' to 'child' and change its description too.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

xv: Add align. pad in struct AdaptorInfo

The (list) 'name' in the struct AdaptorInfo is not padded. Therefore
we've to add an alignment pad after it.

Reference: http://cgit.freedesktop.org/xorg/lib/libXv/tree/src/Xv.c#n250
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=79808

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

dri2: Add notes on lists that should be linked to their length fields but are not.

Reviewed-by: Daniel Martin <consume.noise@gmail.com>

xkb: Fix use of nKTLevels

nLevelsPerType is nLevels long. This continues the work started in
76ca2c0b1527541be59c344118c538ba055ad9d8.

nKTLevels could be used instead of <sumof nLevelsPerType>, but that
change causes a change to libxcb's API.

Reviewed-by: Ran Benita <ran234@gmail.com>
Tested-by: Ran Benita <ran234@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

xprint: Fix <field type="STRING8"> in PrintGetAttributes reply

The reply for PrintGetAttributes contains a field of type STRING8, which
makes no sense as the STRING8 type is only used to denote a string. This
has been changed to <list type="STRING8"> with field "stringLen" as its
length.

Reviewed-by: Peter Harris <pharris@opentext.com>

Remove extraneous spaces between <value> tags

These are the only places where spaces are allowed between <value> tags.
This is not only inconsistent but also complicates the parsing of values
as integers.

Signed-off-by: Uli Schlachter <psychon@znc.in>

Also track directly imported modules in a separate list

This allows the generated header files to only include the
directly referenced header files, with the indirectly referenced header
files included by the directly referenced ones.

Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Uli Schlachter <psychon@znc.in>

xkb: Use <pad align="4" /> in GetMap

There is a bug where lists of size other than one are automatically
aligned by c_client.py. So alignment_pad3 and 4 were aligning twice
and consuming too many bytes.

Rather than change the type to CARD8 and wrap the <op> in yet another
<op> to multiply by two, use the new <pad align> operation.

Reviewed-By: Ran Benita <ran234@gmail.com>
Tested-By: Ran Benita <ran234@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

Support <pad align="n" />

Reviewed-By: Ran Benita <ran234@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

Track pad count at the module level

If the pad count is reset for each bitcase, the names will collide
in the encompasing switch struct.

Reviewed-By: Ran Benita <ran234@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

Release xcb-proto 1.10

Signed-off-by: Uli Schlachter <psychon@znc.in>

Add news for 1.10

Signed-off-by: Julien Cristau <jcristau@debian.org>

xinput: Explicitly specify NotifyMode and NotifyDetail

There is a NotifyMode and a NotifyDetail in both xproto and xinput.  Add
a namespece to the enum references to be clear about which one is to be
used.

Signed-off-by: Peter Harris <pharris@opentext.com>
Signed-off-by: Julien Cristau <jcristau@debian.org>

Add NEWS entries for release 1.9

Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Cristau <jcristau@debian.org>

Rename ge events to GeGeneric events

The new xcb_ge_event_t that libxcb generates from this definition has different
fields than the old, hand-written xcb_ge_event_t. To undo an API break, libxcb
will reintroduce the old hand-written version and mark it as deprecated, so that
everyone can switch to this new version.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=71502
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Cristau <jcristau@debian.org>

sync: VALUETYPE and TESTTYPE are enum

Typoed as mask in e6a246e50e62cbcba33d0e1d2371e69e6e089383

Signed-off-by: Peter Harris <pharris@opentext.com>
Reviewed-By: Daniel Martin <consume.noise@gmail.com>

sync: Add missing namespace for the INT64 struct

Signed-off-by: Peter Harris <pharris@opentext.com>
Reviewed-By: Daniel Martin <consume.noise@gmail.com>

Present: Remove reference to nonexistent enum

Signed-off-by: Peter Harris <pharris@opentext.com>
Reviewed-by: Uli Schlachter <psychon@znc.in>

xkb: Add struct Property as per spec (commented out)

The specification says that Property is a struct of two CountedString16
(name and value). This form can't be handled atm. correctly too. But,
add it as a comment so we've it at hand when we know what to do with it.

All credits go to Ran Benita, as he found this.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

xkb: Comment out intermixed fixed size fields and lists

c_client.py doesn't handle such intermixed fixed size fields and lists
correctly. Therefor comment them out.

With that change the size of the generated request structure for
ListComponents has 8 Bytes which is the same as xkbListComponentsReq
has and the server checks REQUEST_AT_LEAST_SIZE(xkbListComponentsReq).
The same goes for GetKbdByName, which results in a structure of 12
Bytes.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-By: Ran Benita <ran234@gmail.com>

xkb: Comment out the struct Property and {Get, Set}Geometry

This struct has intermixed fixed size fields and lists, which
c_client.py doesn't handle properly. Therefor comment out the struct and
the fields type-referencing it.

As doodads are commented out already and with Property gone the requests
GetGeomtry and SetGeometry become pretty much useless. Therefor comment
them out completly.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-By: Ran Benita <ran234@gmail.com>

xkb: Comment out Doodads

Doodads are known to be broken (*). Comment them out until they can be
handled properly.

(*) One problem is that TextDoodad and LogoDoodad have variadic lists at
the end. But, they're within a union with other fixed size Doodads. A
union doesn't have a hint which field is in use and therefor the code
generator (at least c_client.py) just creates a generic iterator for the
Doodad union, which fails for TextDoodad and LogoDoodad.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-By: Ran Benita <ran234@gmail.com>

Add note on pretty patches and src/.gitattributes

Add a file HACKING describing howto to prettify patches (their hunk
headers) and add src/.gitattributes, which is necessary for that.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-By: Ran Benita <ran234@gmail.com>

xv: Remove obsolete comments in EncodingInfo and QueryEncodings

The padding in EncodingInfo (struct) and QueryEncodings (reply) has been
fixed upstream:
    Xvproto: fix xvEncodingInfo to match actual wire protocol
    http://cgit.freedesktop.org/xorg/proto/videoproto/commit/?id=d970512
and
    Xvproto.h: fix xvQueryEncodingsReply to match actual wire protocol
    http://cgit.freedesktop.org/xorg/proto/videoproto/commit/?id=1a96222

So, we can remove the comments.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Arnaud Fontaine <arnau@debian.org>

xinput: Add XI2 event enum. for event selection

This adds an enumeration 'XIEventMask', which is necessary to select for
XI2 events via the SelectEvent request.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-By: Ran Benita <ran234@gmail.com>

xcbgen: Remove tab from xtypes.py.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=71418

Update to version 1.9

Signed-off-by: Keith Packard <keithp@keithp.com>

Add 'fd' element to fields group, fix present Generic event to validate

This fixes 'make check'

Signed-off-by: Keith Packard <keithp@keithp.com>

Add Present protocol specification

Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Peter Harris <pharris@opentext.com>

Add DRI3

Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Peter Harris <pharris@opentext.com>

Add MIT-SHM AttachFd and CreateSegment requests

Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Peter Harris <pharris@opentext.com>

Add support for file descriptor request fields

These are present in the API, but not present on the wire.

Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>

Define X generic event structure

This is used by libxcb, and was defined explicitly there but it's
cleaner to use the XML specification.

Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Peter Harris <pharris@opentext.com>

Add CARD64/INT64 protocol types

No reason to make people use two 32-bit values when every modern
compiler has native 64-bit objects.

Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>

Move the INT64 types in sync.xml to the sync: namespace

This makes sure we use the crazy sync-specific 64-bit datatype instead
of the global INT64 type

Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Peter Harris <pharris@opentext.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>

XKB: Rewrite AXOption

AXOption can be an <enum> instead of a <union>, which is a much simpler
construct.

Signed-off-by: Peter Harris <pharris@opentext.com>
Reviewed-By: Ran Benita <ran234@gmail.com>

XKB: Fix values of AXFBOpt enum

Signed-off-by: Peter Harris <pharris@opentext.com>
Reviewed-By: Ran Benita <ran234@gmail.com>

xkb: Fix struct ExtensionDeviceNotify

The field ledId is of type CARD16, not CARD8, see XKBproto.h.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-By: Ran Benita <ran234@gmail.com>

xkb: Pad structs CommonDoodad and SALockDeviceBtn

* CommonDoodad is xkbAnyDoodadWireDesc in XKBproto.h. There it has a
  size of 20 like all Doodads:
    #define sz_xkbAnyDoodadWireDesc 20
  Add the necessary 8 bytes of padding.

* SALockDeviceBtn is an Action. All Actions are handled by the server
  and libX11 with sizeof(xkbActionWireDesc) or
  SIZEOF(xkbActionWireDesc). Which results in a size of 8:
    #define sz_xkbActionWireDesc 8 /* in XKBproto.h */
  Add the necessary 3 bytes of padding.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com>

randr: Fix GetProviderProperty reply

The first field has to be 'format' not a pad and 'num_items' is of type
CARD32.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Arnaud Fontaine <arnau@debian.org>

Enforce a bit or value in enum items

Do not allow enum items without an explicit bit or value. The value of
an enum item is known when adding the enum to the xml specification.
There's no reason why this information shouldn't be where it belongs to.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

xv: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

xproto: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

xinput: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

xfixes: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

xevie: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

sync: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

shape: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

screensaver: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

render: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

dri2: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

randr: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

dpms: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

damage: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

composite: Use explicit enum value

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com> (and tested)
Reviewed-by: Josh Triplett <josh@joshtriplett.org>

xkb: Add missing LedClass and BellClass enum values

The protocol encoding says:
http://www.x.org/releases/current/doc/kbproto/xkbproto.html#appD::Common_Types

    KB_LEDCLASSRESULT
        0     KbdFeedbackClass
        4     LedFeedbackClass

    KB_LEDCLASSSPEC
        Encoding same as KB_LEDCLASSRESULT, with the addition of:
        #x0300     XkbDfltXIClass
        #x0500     XkbAllXIClasses

    KB_BELLCLASSRESULT
        0     KbdFeedbackClass
        5     BellFeedbackClass

    KB_BELLCLASSSPEC
        Encoding same as KB_BELLCLASSRESULT, with the addition of:
        #x0300     XkbDfltXIClass

Note that the spec says something slightly different in another section:
http://www.x.org/releases/current/doc/kbproto/xkbproto.html#Common_Types

    KB_LEDCLASSSPEC { KbdFeedbackClass , LedFeedbackClass , DfltXIClass , AllXIClasses , XINone }
    KB_BELLCLASSSPEC    { KbdFeedbackClass , BellFeedbackClass , DfltXIClass , AllXIClasses  }}

From what I could gather, the encoding is the correct one (but I'm not
sure).

Signed-off-by: Ran Benita <ran234@gmail.com>
Reviewed-by: Daniel Martin <consume.noise@gmail.com>

xkb: Add missing 'supported' field to GetNamedIndicator reply

The protocol encoding has it:
http://www.x.org/releases/current/doc/kbproto/xkbproto.html#appD::Requests

    1     CARD8          opcode
    1     15          xkb-opcode
    [...]

    1     1          Reply
    1     CARD8          deviceID
    2     CARD16          sequence number
    [...]
    4     SETofKB_BOOLCTRL          map.ctrls
    1     BOOL          supported
    3               unused

And xkbGetNamedIndicatorReply in XKBproto.h also.

Finally, the server does send it, though it's always TRUE.

Signed-off-by: Ran Benita <ran234@gmail.com>
Reviewed-by: Daniel Martin <consume.noise@gmail.com>

xkb: Unify Overlay1Behavior and Overlay2Behavior

These two structs are used by the server in the exact same way, and are
handled together (see xkb/xkbPrKeyEv.c, XkbKB_Overlay{1,2} handling).
There is no need to use different (and for some reason slightly
different) types for them.

Reviewed-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Ran Benita <ran234@gmail.com>

xkb: Change DfltBtnAbsolute to the value used by the server

The spec says:
     #x02     XkbSA_DfltBtnAbsolute
But /usr/include/X11/extensions/XKB.h has:
     #define XkbSA_DfltBtnAbsolute (1L << 2)
which is what the server sends.

Note: the XKB spec was changed by Daniel Martin to match the Xlib value.
See kbproto commit db07706cb268d5fe1bc38ef5c0a9f49309068b2c.

Reviewed-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Ran Benita <ran234@gmail.com>

xkb: Fix key type map entry field order

In fact, unlike the deleted comment says, both Xlib and the server use
the order as specified in the protocol spec:
http://www.x.org/releases/current/doc/kbproto/xkbproto.html#appD::Requests
(Search for KB_KTMAPENTRY).

Also see struct xkbKTMapEntryWireDesc in
/usr/include/X11/extensions/XKBproto.h

Reviewed-by: Peter Harris <pharris@opentext.com>
Reviewed-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Ran Benita <ran234@gmail.com>

xkb: Work around alignment problems in GetNames and GetMap replies

The basic situation is this: a list of CARD8/CARD16s followed by a list
of CARD16/CARD32s. In the current code, the second list is aligned to
1/2 bytes according the size of the first list. However, in some cases
the second list needs to be aligned to 4 bytes per the xkbproto spec:
http://www.x.org/releases/current/doc/kbproto/xkbproto.html#appD::Requests

XkbGetMap reply (xkb-opcode 8):
[...]
a     LISTofCARD8          actsRtrn.count
p               unused,p=pad(a)
8A     LISTofKB_ACTION          actsRtrn.acts
4B     LISTofKB_SETBEHAVIOR          behaviorsRtrn
v     LISTofSETofKEYMASK          vmodsRtrn
p               unused, p=pad(v)
2E     LISTofKB_SETEXPLICIT          explicitRtrn
p               unused,p=pad(2E)
2M     LISTofKB_KEYMODMAP          modmapRtrn
p               unused, p=pad(2M)
[...]

XkbGetNames reply (xkb-opcode 17):
[...]
l     LISTofCARD8          nLevelsPerType, sum of all elements=L
p               unused, p=pad(l)
[...]

The server and Xlib handle this with calls to XkbPaddedSize(), which is
a good way to see where the extra padding is needed.

Signed-off-by: Ran Benita <ran234@gmail.com>
Reviewed-by: Daniel Martin <consume.noise@gmail.com>

xproto: Fix GContext error references in documentation

The name of the error is GContext, not GC.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-By: Ran Benita <ran234@gmail.com>

xinput: Use BARRIER from xfixes

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

xinput: Unlock uninterpreted lists

The 'uninterpreted_data' lists have been added to pad the structure at
runtime and thereby making lists of such structures iterable.

This is a temporary solution until it is possible to nest a <switch> in
a <struct> correctly.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

xinput: Add XI v2.3

Introduced with v2.3 has been Pointer Barrier support, including a new
request (XIBarrierReleasePointer) and 2 events (Barrier{Hit,Leave}).

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

xinput: Add XI v2.2

Introduced with v2.2 has been Multitouch support and therefor 7 new
events: (Raw)Touch{Begin,Update,End} and TouchOwnership.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

xinput: Add XI v2.1

Introduced with v2.1 has been Raw event support and therefor 5 new
events: Raw{Key,Button}{Press,Release} and RawMotion.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

xinput: Add XI v2.0

Introduced with v2.0 have been various new types, structures, 21
requests and 12 XGE events.

All new requests have the "XI" prefix in the name. This prefix can be
found in the spec too and avoids name clashes with requests found in
prior versions of XI (at least that's (Un)GrabDevice).

All new events are X Generic Events. They've a different header then
"normal" events. Therefor we've to mark them with the attribute 'xge'
set to 'true'.

To prevent another name clash one new structure had to be prefixed,
too: XIDeviceInfo.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

xinput: Add XI v1.5

New:
- Requests
  * ListDeviceProperties
  * ChangeDeviceProperty
  * DeleteDeviceProperty
  * GetDeviceProperty

- Enum
  * PropertyFormat

- Event
  * DevicePropertyNotify

The requests ChangeDeviceProperty and GetDeviceProperty use switches to
return a list depending on the format (8bit, 16bit and 32bit) of the
property value.
We reuse the <bitcase> here. Which is possible, because the format
values don't have equal bits. But, this is rather a hack and must be
changed when the value test (a <case>) for switches is implemented.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

xinput: Cleanup implementation of XI up to v1.4

- Structural changes:
  * indent with spaces only (replaced tabs)
  * indent attributes
  * move structures in front of the first request using it
    (Otherwise c_client.py would generate uncompileable code if we fix
     the uninterpreted lists.)
  * add version-controlled section comments (types, requests, events,
    errors) and move structures into their corresponding section
  * sort events and eventcopys by number

- Additions:
  * add various missing enums and structs (i.e. DeviceChange,
    DeviceControl, DeviceName, ...)
  * uncomment commented out requests
  * linked in various enums into fields (via enum or altenum)

- Other changes:
  * link to the txt specification in the git repo
  * fixed comments for "uninterpreted lists"

- API CHANGES:
  * Request: GetDeviceMotionEvents
    The reply contains a list of events (of coordinates). So, we rename
    num_coords to num_events. The field is called nevents_return in the
    specification. The list was commented out anyways, so the rename
    shouldn't affect that much.
  * Request: SendExtensionEvent
    The list 'events' is a mask, make it a CARD8 like every other mask.
  * Events: FocusIn/Out
    Those ones haven't been named properly. They are called
    DeviceFocusIn/Out in the specification. In XI2 the names FocusIn/Out
    are used to introduce new events and cause name clashes. So, we have
    to rename the existing FocusIn/Out and give them their proper names.
  * Structures: *Feedback{Ctl,State}
    Rename the field id to feedback_id, as it's called in the spec. The
    only lists using those structs are commented out yet. So, this might
    not be called an api break.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>

Add support for X Generic Extension events

With these patches, we are able to mark an XGE event as such and
generate the correct header for it.

XGE events can be found in the X Input Extension v2++.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

sync: Change value list param of CreateAlarm and ChangeAlarm into switch

Values for "Value" and "Delta" fields are 64-bit that couldn't be passed through a regular value list/mask.

Signed-off-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

res: Add ClientIdMask enum

And make use of it in ClientIdSpec.

v2: use <bit/> instead of <value/> in the enum

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

res: Fix and rename list in ClientIdValue

The list is a CARD32 and rename the field to "value" to match the
specification.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

randr: Add provider object support (RandR v1.4)

With RandR 1.4 provider objects have been introduced. "A provider object
represents a GPU or virtual device providing services to the X server."

To handle provider objects the following requests have been added:
    - GetProviders
    - GetProviderInfo
    - SetProviderOffloadSink
    - SetProviderOutputSource
    - ListProviderProperties
    - QueryProviderProperty
    - ConfigureProviderProperty
    - ChangeProviderProperty
    - DeleteProviderProperty
    - GetProviderProperty

And the Notify event got 3 new subcodes:
    - ProviderChange
    - ProviderProperty
    - ResourceChange

v2:
- add missing name field to GetProviderInfo
v3:
- changed ProviderCapability enum
  o replaced <value> with <bit>
  o removed None
- replaced "alignment pad list" with commented out <pad align="4"/> to
  be prepared for future enhancements
- add brief description to commit msg about additions

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

xtest: Version bump 2.1..2.2

The implemented and upstream version is 2.2.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

glx: Version bump 1.3..1.4

Add the BufferSwapComplete (v2) event.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

composite: Version bump 0.3..0.4

The only difference between version 0.3 and 0.4 is that the clipping
semantics have been redefined. That's just an internal change, no
structure has to be modified.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

xfixes: Rename enum DirectionMask to BarrierDirec.

Rename the enum 'DirectionMask' to 'BarrierDirections' to match the
specification.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

XKB: Correct enum to mask

Changes fields with enums EventType, MapPart, or BoolCtrl to use them as
masks instead.

screensaver: Use enum and mask attributes

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>

screensaver: Remove wrong fields from Notify event

This fixes bug #63158:
    screensaver notify event specification is incorrect
    https://bugs.freedesktop.org/show_bug.cgi?id=63158

As stated in the bug report the response_type and sequence field will be
added automatically while parsing the specification. Additionally, the
pad field is wrong.

Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>