platform/upstream/libxkbcommon.git
14 years agoDrop Bool type
Kristian Høgsberg [Fri, 2 Jul 2010 19:05:21 +0000 (15:05 -0400)]
Drop Bool type

It's a nightmare trying to avoid double definitions.

14 years agoGet rid of a few unused #defines in public header
Kristian Høgsberg [Fri, 2 Jul 2010 16:26:43 +0000 (12:26 -0400)]
Get rid of a few unused #defines in public header

14 years agoRename XkbRMLVOSet to struct xkb_rule_names
Kristian Høgsberg [Fri, 2 Jul 2010 16:14:03 +0000 (12:14 -0400)]
Rename XkbRMLVOSet to struct xkb_rule_names

14 years agoMove private structs and defines to private headers
Kristian Høgsberg [Fri, 2 Jul 2010 16:10:05 +0000 (12:10 -0400)]
Move private structs and defines to private headers

14 years agoRename public entry points to lowercase and underscore
Kristian Høgsberg [Fri, 2 Jul 2010 15:50:01 +0000 (11:50 -0400)]
Rename public entry points to lowercase and underscore

14 years agoCompile with -fvisibility=hidden when possible
Kristian Høgsberg [Fri, 2 Jul 2010 15:43:56 +0000 (11:43 -0400)]
Compile with -fvisibility=hidden when possible

14 years agoPull in enough structs and defines from XKBstr.h to only need XKB.h
Kristian Høgsberg [Thu, 1 Jul 2010 18:35:24 +0000 (14:35 -0400)]
Pull in enough structs and defines from XKBstr.h to only need XKB.h

We want to move away from sharing implementation structs and let libX11
and libxkbcommon use each their own set of structs.

14 years agoDrop Xmd.h include now that we're using uint32_t
Kristian Høgsberg [Wed, 30 Jun 2010 21:32:54 +0000 (17:32 -0400)]
Drop Xmd.h include now that we're using uint32_t

14 years agoFix warning from CARD32 -> uint32_t conversion
Kristian Høgsberg [Wed, 30 Jun 2010 21:31:34 +0000 (17:31 -0400)]
Fix warning from CARD32 -> uint32_t conversion

14 years agoConstify XkbcAtomText()
Kristian Høgsberg [Wed, 30 Jun 2010 21:20:56 +0000 (17:20 -0400)]
Constify XkbcAtomText()

Atoms aren't mutable and this lets us put tbGetBuffer() back in the box.

14 years agoMake XkbcInitAtoms() call optional
Kristian Høgsberg [Wed, 30 Jun 2010 21:13:21 +0000 (17:13 -0400)]
Make XkbcInitAtoms() call optional

14 years agoDrop more malloc/free wrappers
Kristian Høgsberg [Wed, 30 Jun 2010 20:56:24 +0000 (16:56 -0400)]
Drop more malloc/free wrappers

14 years agoUse the right action structs
Kristian Høgsberg [Wed, 30 Jun 2010 20:33:25 +0000 (16:33 -0400)]
Use the right action structs

We were casting the union members to the wrong action structs (from kbproto).

14 years agoCopy over missing action structs from kbproto
Kristian Høgsberg [Wed, 30 Jun 2010 19:48:39 +0000 (15:48 -0400)]
Copy over missing action structs from kbproto

14 years agoRename Xkbc*Action to struct xkb_*_action
Kristian Høgsberg [Wed, 30 Jun 2010 17:31:21 +0000 (13:31 -0400)]
Rename Xkbc*Action to struct xkb_*_action

14 years agoDrop malloc wrappers
Kristian Høgsberg [Mon, 28 Jun 2010 10:58:01 +0000 (06:58 -0400)]
Drop malloc wrappers

14 years agoDrop CARD32 and Opaque types
Kristian Høgsberg [Mon, 28 Jun 2010 10:50:12 +0000 (06:50 -0400)]
Drop CARD32 and Opaque types

14 years agoCopy and duplicate XkbModsRec and XkbKTMapEntryRec
Daniel Stone [Wed, 23 Jun 2010 15:25:10 +0000 (16:25 +0100)]
Copy and duplicate XkbModsRec and XkbKTMapEntryRec

Copy these types in so we can extend the vmod size.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agofix mod size confusion
Daniel Stone [Wed, 23 Jun 2010 15:25:02 +0000 (16:25 +0100)]
fix mod size confusion

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoAdd really rudimentary rules caching support
Daniel Stone [Tue, 22 Jun 2010 14:54:52 +0000 (15:54 +0100)]
Add really rudimentary rules caching support

Keep the parsed form of the last-used rules file around, and reuse that
if we get asked for the same ruleset.  If not, bin it and cache the
other one.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoUse CARD32 instead of Atom, move geom headers in
Daniel Stone [Mon, 21 Jun 2010 13:28:34 +0000 (14:28 +0100)]
Use CARD32 instead of Atom, move geom headers in

Use CARD32 instead of Atom/KeySym/et al to avoid type size confusion
between server and non-server code; relatedly, move the geometry headers
in from kbproto, so every non-simple type (i.e. structs containing
nothing more than basic types) is now copied into xkbcommon.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoInterp: More lazy keysym resolution
Daniel Stone [Mon, 21 Jun 2010 13:27:58 +0000 (14:27 +0100)]
Interp: More lazy keysym resolution

Resolve the keysyms when we create an InterpDef, rather than directly
in the parser.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoAllow external atom databases
Daniel Stone [Mon, 21 Jun 2010 13:22:26 +0000 (14:22 +0100)]
Allow external atom databases

Allow people to plug in an external atom database (e.g. the X server's),
so we don't have to migrate our own atoms over later.  We are a bit
over-keen on atoms at the moment, so it does pollute the atom database a
bit though.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoCopy in XkbCompatMapRec and XkbSymInterpretRec
Daniel Stone [Thu, 17 Jun 2010 04:56:08 +0000 (05:56 +0100)]
Copy in XkbCompatMapRec and XkbSymInterpretRec

These contain actions, so transition them ahead to XkbcAction and move
them into XKBcommon.h.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoUse CARD32 instead of Atom, drag in XkbClientMapRec
Daniel Stone [Thu, 17 Jun 2010 02:16:09 +0000 (03:16 +0100)]
Use CARD32 instead of Atom, drag in XkbClientMapRec

On 64-bit architectures, XID varies in size between the server (always
32 bits), and non-server (always unsigned long) for some inexplicable
reason.  Use CARD32 instead to avoid this horrible trap.

This involves dragging in XkbClientMapRec so we don't get stuck in the
KeySym trap.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoRegroup actions into current vs. deprecated, resize vmod
Daniel Stone [Wed, 16 Jun 2010 23:51:49 +0000 (00:51 +0100)]
Regroup actions into current vs. deprecated, resize vmod

Use Xkbc* for all our actions that we intend to keep around, and Xkb*
for deprecated ones we can hopefully get rid of, at least internally.

While we're at it, make vmods be a uint32_t.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: Don't malloc() and free() most scanned tokens
Daniel Stone [Tue, 15 Jun 2010 18:43:14 +0000 (19:43 +0100)]
xkbcomp: Don't malloc() and free() most scanned tokens

Use a constant buffer instead.  Sigh.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoFix compilation with DEBUG
Daniel Stone [Tue, 15 Jun 2010 18:40:31 +0000 (19:40 +0100)]
Fix compilation with DEBUG

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: Use fread() instead of getc()
Daniel Stone [Tue, 15 Jun 2010 18:39:15 +0000 (19:39 +0100)]
xkbcomp: Use fread() instead of getc()

So, it turns out that if you're parsing a fairly large amount of data,
using getc() to get all the input rather than, say, read(), is some kind
of remarkably daft and unperformant idea.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: Lazy keysym parsing (avoid XStringToKeysym)
Daniel Stone [Tue, 15 Jun 2010 18:38:16 +0000 (19:38 +0100)]
xkbcomp: Lazy keysym parsing (avoid XStringToKeysym)

Instead of calling XStringToKeysym on every keysym we parse, store it as
a string until we need to store it in an actual keymap.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoAtoms: Avoid allocations in XkbAtomText()
Daniel Stone [Tue, 15 Jun 2010 18:30:30 +0000 (19:30 +0100)]
Atoms: Avoid allocations in XkbAtomText()

XkbAtomGetString() returns a freshly-allocated string, whereas
XkbAtomText() returns the same in a temporary buffer.  XkbAtomText used
to call XkbAtomGetString() and then free the result, which seems quite
spectacularly pointless when you think about it.  Shuffle the atom code
around so we don't have to allocate for XkbAtomText().

This changes semantics slightly wrt non-printable characters, but I
haven't been able to see any effect so far.  And it may well be ever so
slightly quicker.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoconfigure.ac: Enable automake silent rules by default
Daniel Stone [Tue, 15 Jun 2010 15:43:10 +0000 (16:43 +0100)]
configure.ac: Enable automake silent rules by default

This allows us to spot warnings much more easily.  Also, drop the
now-deprecated AM_MAINTAINER_MODE.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoExpose tbGetBuffer to the rest of libxkbcommon
Daniel Stone [Tue, 15 Jun 2010 15:42:06 +0000 (16:42 +0100)]
Expose tbGetBuffer to the rest of libxkbcommon

This will let us implement XkbAtomGetString on top of XkbAtomText,
instead of having the latter get a duplicated string, dump it into a
temporary buffer, and subsequently free it (sigh).

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: keycodes: Silence gcc warnings
Daniel Stone [Tue, 15 Jun 2010 15:41:10 +0000 (16:41 +0100)]
xkbcomp: keycodes: Silence gcc warnings

It was right to warn us: some of these were actually bogus.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: geometry: Plug numerous atom text & expr leaks
Daniel Stone [Tue, 15 Jun 2010 14:27:12 +0000 (15:27 +0100)]
xkbcomp: geometry: Plug numerous atom text & expr leaks

This code is absolute misery.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: Atom text and expr leak fixes part #973
Daniel Stone [Tue, 15 Jun 2010 14:25:40 +0000 (15:25 +0100)]
xkbcomp: Atom text and expr leak fixes part #973

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: Don't leak atom text and string exprs, again
Daniel Stone [Tue, 15 Jun 2010 14:23:23 +0000 (15:23 +0100)]
xkbcomp: Don't leak atom text and string exprs, again

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: expr: Use XkbcAtomText instead of GetString
Daniel Stone [Tue, 15 Jun 2010 14:22:05 +0000 (15:22 +0100)]
xkbcomp: expr: Use XkbcAtomText instead of GetString

This returns us a temporary string, rather than leaking the atom.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: Don't leak atom text and string exprs
Daniel Stone [Tue, 15 Jun 2010 14:20:32 +0000 (15:20 +0100)]
xkbcomp: Don't leak atom text and string exprs

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: vmod: Don't get and immediately intern atoms
Daniel Stone [Tue, 15 Jun 2010 14:14:58 +0000 (15:14 +0100)]
xkbcomp: vmod: Don't get and immediately intern atoms

XkbcInternAtom(XkbcAtomGetString(atom)) has to be the most spectacularly
broken antipattern I've yet seen.  Just compare the atoms directly.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoxkbcomp: Replace open-coded strdup
Daniel Stone [Tue, 15 Jun 2010 14:13:56 +0000 (15:13 +0100)]
xkbcomp: Replace open-coded strdup

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoAdd testcase for XkbcCanonicaliseComponents
Daniel Stone [Sat, 25 Apr 2009 08:15:02 +0000 (18:15 +1000)]
Add testcase for XkbcCanonicaliseComponents

Should be reasonably thorough; the shell part is, however, quite
unpleasant.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoAdd XkbcCanonicaliseComponents
Daniel Stone [Sat, 25 Apr 2009 08:13:52 +0000 (18:13 +1000)]
Add XkbcCanonicaliseComponents

Canonicalises two sets of components (new and old), e.g.:
new: +bar
old: foo
result: foo+bar

This is required as part of the spec, so clients can reuse part of the
device's old keymap.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoXKBcommon.h: Make header self-contained
Daniel Stone [Sat, 25 Apr 2009 08:13:24 +0000 (18:13 +1000)]
XKBcommon.h: Make header self-contained

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoKeysym: Add apallingly bad API documentation for keysym <-> string
Daniel Stone [Sat, 25 Apr 2009 04:55:59 +0000 (14:55 +1000)]
Keysym: Add apallingly bad API documentation for keysym <-> string

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoMake keysym <-> string conversion public API
Daniel Stone [Sat, 25 Apr 2009 04:52:51 +0000 (14:52 +1000)]
Make keysym <-> string conversion public API

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoXKB: Text: Use keysym <-> string conversion from keysym.c
Daniel Stone [Sat, 25 Apr 2009 04:51:46 +0000 (14:51 +1000)]
XKB: Text: Use keysym <-> string conversion from keysym.c

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoTest: Keysym: Add tests for new keysym <-> string conversions
Daniel Stone [Sat, 25 Apr 2009 04:51:16 +0000 (14:51 +1000)]
Test: Keysym: Add tests for new keysym <-> string conversions

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
14 years agoKeysym: Fix conversion for Unicode and bare numbers
Daniel Stone [Sat, 25 Apr 2009 04:42:42 +0000 (14:42 +1000)]
Keysym: Fix conversion for Unicode and bare numbers

Ensure that Unicode string representations are accepted and turned into
numbers, as well as hexadecimal numbers in 0xabcd1234 form; unknown
keysyms are output as 0xabcd1234 in string form.

This also ensures that strings are never returned malloc()ed.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
15 years agoKeySym: Actually handle NoSymbol
Daniel Stone [Sat, 25 Apr 2009 04:32:04 +0000 (14:32 +1000)]
KeySym: Actually handle NoSymbol

Add NoSymbol into the keysym table, so keysym <-> string conversion works for
that, too; also eliminate special-casing of VoidSymbol.

This will require special-casing in libX11 to preserve its API.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
15 years agotest: Fix srcdir != objdir build and test failures
Daniel Stone [Sat, 25 Apr 2009 04:29:23 +0000 (14:29 +1000)]
test: Fix srcdir != objdir build and test failures

filecomp would fail because it couldn't find the input files, after
compilation failed due to missing includes.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
15 years agolisting: Drop unused deviceSpec argument
Dan Nicholson [Wed, 22 Apr 2009 06:22:25 +0000 (23:22 -0700)]
listing: Drop unused deviceSpec argument

The server might care about this, but ListComponents is just reading xkb
files.

15 years agoxkbcomp: Another fairly major rewrite of the listing mode
Dan Nicholson [Mon, 20 Apr 2009 13:40:34 +0000 (06:40 -0700)]
xkbcomp: Another fairly major rewrite of the listing mode

The listing code in xkbcomp had been setup to allocate a set of buffers
with file paths and then later parse through them to find which maps were
needed.

All the allocation (with the additional allocation for the components
list) was making it really slow, so this patch makes everything simpler
by just generating the components list as we walk the directory tree.

15 years agoRewrite listing code to generate XkbComponentListPtr
Dan Nicholson [Sun, 19 Apr 2009 23:30:36 +0000 (16:30 -0700)]
Rewrite listing code to generate XkbComponentListPtr

In xkbcomp, the listing code printed out the xkb files to stdout and the
caller would parse the output. Here, we can just generate a
XkbComponentListPtr and pass it back.

This should be a series of smaller commits, but there was no clean way to
do it since it's basically a complete rewrite except for the core map
matching logic.

A lot of code used for special printing modes in xkbcomp has been
shedded. Callers can massage the output as they please.

15 years agoxkbcomp: Kill unused listing code and make functions static
Dan Nicholson [Fri, 17 Apr 2009 04:47:28 +0000 (21:47 -0700)]
xkbcomp: Kill unused listing code and make functions static

Much of this was only called from xkbcomp and not used to generate
listings for the server.

15 years agoScale back global variables
Dan Nicholson [Fri, 17 Apr 2009 03:01:13 +0000 (20:01 -0700)]
Scale back global variables

Some of the global variables were either not used anymore or only used
in the listing code.

15 years agoAdd XkbNameMatchesPattern implementation from xkbfile
Dan Nicholson [Sun, 12 Apr 2009 19:30:12 +0000 (12:30 -0700)]
Add XkbNameMatchesPattern implementation from xkbfile

The xkbcomp listing code matches a glob type pattern against installed
xkb files. This adds a Xkbc implementation of the pattern matching code.

15 years agoxkbcomp: Fix use of removed macros
Dan Nicholson [Sun, 12 Apr 2009 19:15:05 +0000 (12:15 -0700)]
xkbcomp: Fix use of removed macros

15 years agoRevert "xkbcomp: Remove listing sources"
Dan Nicholson [Sun, 12 Apr 2009 18:13:52 +0000 (11:13 -0700)]
Revert "xkbcomp: Remove listing sources"

This reverts commit c4c9e36fbf4019d802323d9053f54cbfa6d3d5e5. It turns
out that the listing code is used to support the X_kbListComponents
request (via XkbListComponents).

This will have to be refactored into some reasonable interface instead
of the current usage where the server reads xkbcomp stdout. Gross.

15 years agomisc: Coding-style cleanup
Dan Nicholson [Sun, 12 Apr 2009 19:14:36 +0000 (12:14 -0700)]
misc: Coding-style cleanup

15 years agoUse xkbpath API for locating rules
Dan Nicholson [Sat, 11 Apr 2009 17:35:44 +0000 (10:35 -0700)]
Use xkbpath API for locating rules

Instead of hardcoding the XKB base directory when searching for rules in
the xkbcomp code, we can extend the xkbpath API to cover rules and reuse
it. That will make it more convenient if it's ever exposed so people can
set their XKB search paths in a reasonable way.

15 years agoProgram and files for testing CompileKeymapFromFile
Dan Nicholson [Fri, 10 Apr 2009 19:33:31 +0000 (12:33 -0700)]
Program and files for testing CompileKeymapFromFile

A few simple test cases for verifying the operation of parsing a keymap
file and compiling a keyboard description from it.

15 years agoAdd interface to compile keyboard description from keymap files
Dan Nicholson [Fri, 10 Apr 2009 19:25:51 +0000 (12:25 -0700)]
Add interface to compile keyboard description from keymap files

We need to support generating a keyboard description from a keymap file
because there are just some cases where RMLVO or ktcsg is not enough.

The map choosing logic has been refactored into its own function and now
supports choosing a named or default keymap.

15 years agotest: Remove extraneous includes
Dan Nicholson [Fri, 10 Apr 2009 03:52:26 +0000 (20:52 -0700)]
test: Remove extraneous includes

These programs don't actually use the internal API.

15 years agoMake XkbcInitAtoms externally accessible
Dan Nicholson [Thu, 9 Apr 2009 21:29:32 +0000 (14:29 -0700)]
Make XkbcInitAtoms externally accessible

Applications (like the server) need to initialize the atoms system
before using the rest of the library. Maybe it should just init itself
implicitly.

15 years agoConstify public API
Dan Nicholson [Thu, 9 Apr 2009 00:59:15 +0000 (17:59 -0700)]
Constify public API

There's no reason the arguments can't be const.

15 years agoChange CompileKeymapFromRules to take XkbRMLVOSet
Dan Nicholson [Thu, 9 Apr 2009 00:54:55 +0000 (17:54 -0700)]
Change CompileKeymapFromRules to take XkbRMLVOSet

Let's use a nice interface now that it's available from XKBrulescommon.h.

15 years agoRemove all non-public API from XKBcommon.h header
Dan Nicholson [Wed, 8 Apr 2009 14:46:25 +0000 (07:46 -0700)]
Remove all non-public API from XKBcommon.h header

The noble intention was to expose all the new API and new generic types
in the split out kbproto headers through XKBcommon.h. It turns out that
would be a massive amount of work in the server. Someday, but first just
wedging in XkbCompileKeymap* would be good.

Most of the API is in new internal xkb*.h headers. In order to allow the
XKBcommon.h header to be used from the server, we can't pull in other
headers from kbproto since the server has its own copies. However, types
that are different (XkbDescRec, XkbAction) still have Xkbc equivalents
here, and I think they should be used in the server.

15 years agoRemove KS macros available in XKBfilecommon.h now
Dan Nicholson [Wed, 8 Apr 2009 02:56:41 +0000 (19:56 -0700)]
Remove KS macros available in XKBfilecommon.h now

A couple of the XkbcKS* macros become redundant with XKBfilecommon.h
split out in kbproto.

15 years agoRequire strdup and remove utils wrapper
Dan Nicholson [Mon, 6 Apr 2009 03:27:35 +0000 (20:27 -0700)]
Require strdup and remove utils wrapper

This kills a couple warnings from using the uStringDup wrapper. If you
don't have strdup on your platform, you have bigger issues.

15 years agoAdd action datatypes as defined in the server
Dan Nicholson [Sun, 5 Apr 2009 22:15:20 +0000 (15:15 -0700)]
Add action datatypes as defined in the server

Some of the XkbAction types are defined differently in the server, so we
add those to XKBcommon.h and use them here like XkbcDescPtr. We'll have
to deal with the impedance mismatch on the client side later.

15 years agoCoding style cleanup on key allocation functions
Dan Nicholson [Sun, 5 Apr 2009 16:30:15 +0000 (09:30 -0700)]
Coding style cleanup on key allocation functions

15 years agotest: Exercise compiling from components
Dan Nicholson [Sat, 4 Apr 2009 19:54:44 +0000 (12:54 -0700)]
test: Exercise compiling from components

This could probably use a lot more real world test cases, but it does
the job for now.

15 years agoxkbcomp: Don't say we're exiting when we're not
Dan Nicholson [Sat, 4 Apr 2009 19:50:27 +0000 (12:50 -0700)]
xkbcomp: Don't say we're exiting when we're not

15 years agoxkbcomp: Give keymap file empty name if none supplied
Dan Nicholson [Sat, 4 Apr 2009 19:31:55 +0000 (12:31 -0700)]
xkbcomp: Give keymap file empty name if none supplied

Ensure that the keymap file topName is not NULL since it gets blindly
dereferenced later.

15 years agoxkbcomp: Ensure at least keycodes supplied in components
Dan Nicholson [Sat, 4 Apr 2009 19:21:05 +0000 (12:21 -0700)]
xkbcomp: Ensure at least keycodes supplied in components

It seems that at least a non-empty keycodes component is required to
compile a XkbcDescPtr.

15 years agoxkbcomp: Don't dereference NULL VarDefsPtr
Dan Nicholson [Sat, 4 Apr 2009 19:17:50 +0000 (12:17 -0700)]
xkbcomp: Don't dereference NULL VarDefsPtr

15 years agoxkbcomp: Refactor keymap file generator
Dan Nicholson [Sat, 4 Apr 2009 19:16:04 +0000 (12:16 -0700)]
xkbcomp: Refactor keymap file generator

15 years agotest: Minor refactor to allow $EXEEXT in program names
Dan Nicholson [Sat, 4 Apr 2009 18:04:35 +0000 (11:04 -0700)]
test: Minor refactor to allow $EXEEXT in program names

15 years agotest: Add logging and some intentionally failing cases
Dan Nicholson [Sat, 4 Apr 2009 17:52:23 +0000 (10:52 -0700)]
test: Add logging and some intentionally failing cases

We want to log the output of the tests rather than letting them go to
stderr. This allows tests we expect to fail to be run.

15 years agoxkbcomp: Generating components requires rules _and_ layout
Dan Nicholson [Sat, 4 Apr 2009 17:02:39 +0000 (10:02 -0700)]
xkbcomp: Generating components requires rules _and_ layout

This came from trial and error, but it seems that you can generate a
valid keymap with only rules and layout.

15 years agoxkbcomp: Fail when converting rules to components returns an error
Dan Nicholson [Sat, 4 Apr 2009 16:51:33 +0000 (09:51 -0700)]
xkbcomp: Fail when converting rules to components returns an error

XkbcRF_GetComponents was returning an error but leaving the generated
components alone. This ensures that the broken XkbComponentNamesPtr is
freed and the error is passed up to the caller.

15 years agoxkbcomp: Cleanup error messages
Dan Nicholson [Sat, 4 Apr 2009 16:48:59 +0000 (09:48 -0700)]
xkbcomp: Cleanup error messages

15 years agoxkbcomp: Ensure user has supplied a rules name
Dan Nicholson [Sat, 4 Apr 2009 16:46:20 +0000 (09:46 -0700)]
xkbcomp: Ensure user has supplied a rules name

15 years agorulescomp: Remove unneeded API
Dan Nicholson [Sat, 4 Apr 2009 16:37:53 +0000 (09:37 -0700)]
rulescomp: Remove unneeded API

15 years agoRefactor xkbpath so that it implicitly initializes
Dan Nicholson [Sat, 4 Apr 2009 16:29:43 +0000 (09:29 -0700)]
Refactor xkbpath so that it implicitly initializes

Instead of requiring the user to call XkbInitIncludePath() and
XkbAddDefaultDirectoriesToPath(), all the path entry points now implicitly
initialize the path. When initializing, the default directories are added
so it's useful.

This provides normal operation without exposing the xkbpath API. That
might happen later to allow apps to edit the XKB search path.

15 years agoRemove trailing spaces in source files
Dan Nicholson [Sat, 4 Apr 2009 16:19:51 +0000 (09:19 -0700)]
Remove trailing spaces in source files

15 years agoTest compiler to simulate xkbcomp usage
Dan Nicholson [Sat, 4 Apr 2009 16:14:20 +0000 (09:14 -0700)]
Test compiler to simulate xkbcomp usage

Added a test program, rulescomp, which takes a RMLVO set and generates a
XkbcDescPtr. This is essentially what the xserver will do, except that we
still need to access some xkbcomp internal API to make it work.

15 years agoFix cleanup error in names allocation
Dan Nicholson [Sat, 4 Apr 2009 15:52:17 +0000 (08:52 -0700)]
Fix cleanup error in names allocation

A conditional got reversed during the coding style cleanup.

15 years agoxkbcomp: Ignore generated parser
Dan Nicholson [Sat, 4 Apr 2009 04:18:58 +0000 (21:18 -0700)]
xkbcomp: Ignore generated parser

15 years agoxkbcomp: Don't segfault if the debug files aren't set
Dan Nicholson [Sat, 4 Apr 2009 04:16:56 +0000 (21:16 -0700)]
xkbcomp: Don't segfault if the debug files aren't set

Another interface that needs to be cleaned up.

15 years agoxkbcomp: Declare global debugging vars
Dan Nicholson [Sat, 4 Apr 2009 03:34:41 +0000 (20:34 -0700)]
xkbcomp: Declare global debugging vars

These will need to be sanitized at some point, but for now let's just get
the code running.

15 years agoxkbcomp: Back out strdup warning cleanup
Dan Nicholson [Sat, 4 Apr 2009 03:33:01 +0000 (20:33 -0700)]
xkbcomp: Back out strdup warning cleanup

There are some spots in the code that use strdup without checking the
argument, so we need this to not segfault. Cleanup later...

15 years agoFix install path for XKBcommon.h
Dan Nicholson [Sat, 4 Apr 2009 02:35:01 +0000 (19:35 -0700)]
Fix install path for XKBcommon.h

15 years agoDrop keysym.h pollution from XKBcommon.h
Dan Nicholson [Thu, 2 Apr 2009 13:37:16 +0000 (06:37 -0700)]
Drop keysym.h pollution from XKBcommon.h

There's really no need to pull this into a public header.

15 years agoAdd xkbcommon pkg-config file
Dan Nicholson [Wed, 1 Apr 2009 13:22:06 +0000 (06:22 -0700)]
Add xkbcommon pkg-config file

15 years agoxkbcomp: Clean up utils prototypes
Dan Nicholson [Wed, 1 Apr 2009 02:30:44 +0000 (19:30 -0700)]
xkbcomp: Clean up utils prototypes

Includes using a macro for the gcc format attributes instead of repeated
ifdef's.

15 years agoxkbcomp: Remove duplicated macros
Dan Nicholson [Tue, 31 Mar 2009 14:21:20 +0000 (07:21 -0700)]
xkbcomp: Remove duplicated macros

15 years agoxkbcomp: Kill warnings about address always evaluating as true
Dan Nicholson [Tue, 31 Mar 2009 14:09:47 +0000 (07:09 -0700)]
xkbcomp: Kill warnings about address always evaluating as true

15 years agoMerge commit 'tilt/master'
Dan Nicholson [Tue, 31 Mar 2009 13:57:06 +0000 (06:57 -0700)]
Merge commit 'tilt/master'