Sven Verdoolaege [Thu, 20 Oct 2011 18:18:57 +0000 (20:18 +0200)]
isl_tab_pip.c: set_row_cst_to_div: don't assume coefficient of div is zero
The function is supposed to increment the coefficient of the given
div by one, but instead it would simply set the coefficient to one.
This is fine if the div is new and therefore has a zero coefficient,
but it does not work if the div already appeared in the constraint.
Reported-by: Armin Größlinger <armin.groesslinger@uni-passau.de>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 20 Oct 2011 18:17:41 +0000 (20:17 +0200)]
isl_tab_pip.c: fix typo in comment
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 19 Oct 2011 08:23:23 +0000 (10:23 +0200)]
remove undocumented isl_basic_{set,map}_foreach_lex{min,max}
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 19 Oct 2011 08:14:48 +0000 (10:14 +0200)]
isl_test: test isl_basic_map_lexmin_pw_multi_aff
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 16 Oct 2011 14:00:38 +0000 (16:00 +0200)]
add isl_union_map_gist_domain
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 16 Oct 2011 13:49:40 +0000 (15:49 +0200)]
add isl_map_gist_domain
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 16 Oct 2011 13:07:44 +0000 (15:07 +0200)]
add isl_union_map_gist_params
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 16 Oct 2011 08:55:57 +0000 (10:55 +0200)]
isl_union_map_intersect_domain: special case intersection with parameter domain
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 16 Oct 2011 08:55:40 +0000 (10:55 +0200)]
isl_union_set_intersect: special case intersection with parameter domain
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 16 Oct 2011 08:54:22 +0000 (10:54 +0200)]
add isl_union_map_intersect_params
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sat, 15 Oct 2011 08:45:08 +0000 (10:45 +0200)]
add isl_union_set_is_params
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Fri, 14 Oct 2011 10:10:13 +0000 (12:10 +0200)]
isl_point_get_coordinate: check validity of "pos" argument
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 13 Oct 2011 14:26:02 +0000 (16:26 +0200)]
interface/extract_interface.cc: adjust to rename of Diagnostic in clang
Reported-by: Tobias Grosser <tobias@grosser.es>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 13 Oct 2011 10:16:45 +0000 (12:16 +0200)]
rename isl_pw_aff_add to isl_pw_aff_union_add
As pointed out by Andreas Kloeckner, isl_pw_aff_add behaves in a way
that is inconsistent with other functions defined over pairs of
isl_pw_affs in that it produces an isl_pw_aff that is defined on
the union of the definition domains of its arguments.
We therefore rename isl_pw_aff_add to isl_pw_aff_union_add and
add a new isl_pw_aff_add that produces a result that is only
defined on the intersection of the definition domains of its arguments.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 13 Oct 2011 09:54:16 +0000 (11:54 +0200)]
reimplement pw_aff_mul on top of a generic isl_pw_*_on_shared_domain
This new function will be useful for implementing other similar functionality.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 12 Oct 2011 09:52:59 +0000 (11:52 +0200)]
isl_map_intersect: require arguments to live in the same space
Before, we would allow one of the arguments to be a parameter domain.
As pointed out by Andreas Kloeckner, this breaks some invariants
in other parts of the code, notably isl_pw_*_intersect_domain.
Since we have an isl_map_intersect_params now, we no longer need
this special case, so we remove it.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 12 Oct 2011 09:49:42 +0000 (11:49 +0200)]
isl_input.c: read_tuple: use isl_map_intersect_params where needed
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Mon, 10 Oct 2011 22:27:35 +0000 (00:27 +0200)]
Merge branch 'maint'
Andreas Kloeckner [Mon, 10 Oct 2011 22:02:18 +0000 (18:02 -0400)]
Insert missing error check into isl_pw_*_intersect_domain
Signed-off-by: Andreas Kloeckner <inform@tiker.net>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Andreas Kloeckner [Mon, 10 Oct 2011 12:09:50 +0000 (08:09 -0400)]
Run isl_basic_map_finalize after isl_{basic_,}map_eliminate
Signed-off-by: Andreas Kloeckner <inform@tiker.net>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sat, 8 Oct 2011 14:23:59 +0000 (16:23 +0200)]
isl_stream_read_map: handle nested parenthesized expressions
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 6 Oct 2011 21:19:43 +0000 (23:19 +0200)]
isl_basic_{set,map}_partial_lex{min,max}_pw_multi_aff: normalize isl_affs
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 6 Oct 2011 21:18:51 +0000 (23:18 +0200)]
add isl_aff_normalize
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 2 Aug 2011 10:58:48 +0000 (12:58 +0200)]
add extract_interface for constructing interfaces to other languages
Currently, we only support the generation of a Python interface.
In particular, a ctypes based interface.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 21 Jul 2010 07:55:23 +0000 (09:55 +0200)]
mark some data structures and functions for export to language bindings
We prefer to only include function in interfaces to other languages
that have been explicitly marked for export.
In particular, we do not want to add functions that are obsolescent
and are only kept for backward compatibility, or functions that
only make sense for a C interface, or functions that are only meant
for other libraries such as barvinok that are tightly coupled to isl.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 21 Jul 2010 08:00:38 +0000 (10:00 +0200)]
add isl_*_to_str
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 4 Sep 2011 08:47:36 +0000 (10:47 +0200)]
add isl_union_pw_qpolynomial_read_from_str
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 8 Sep 2011 20:08:18 +0000 (22:08 +0200)]
add isl_multi_aff_lift
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 8 Sep 2011 20:08:03 +0000 (22:08 +0200)]
add isl_local_space_lifting
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 8 Sep 2011 20:08:03 +0000 (22:08 +0200)]
add isl_local_space_lift
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 8 Sep 2011 12:32:58 +0000 (14:32 +0200)]
add isl_local_space_intersect
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 28 Aug 2011 16:15:39 +0000 (18:15 +0200)]
add isl_multi_aff_read_from_str
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 21 Sep 2011 12:23:46 +0000 (14:23 +0200)]
add isl_pw_aff_read_from_str
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 28 Aug 2011 15:08:00 +0000 (17:08 +0200)]
add isl_aff_read_from_str
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 28 Aug 2011 12:18:46 +0000 (14:18 +0200)]
add isl_pw_multi_aff_read_from_str
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 16 Aug 2011 16:46:37 +0000 (18:46 +0200)]
add isl_pw_multi_aff_from_map
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 16 Aug 2011 16:46:25 +0000 (18:46 +0200)]
add isl_basic_map_lexmin_pw_multi_aff
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 1 Sep 2011 10:46:39 +0000 (12:46 +0200)]
pip: optionally use isl_basic_set_partial_lexmin_pw_multi_aff
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sat, 13 Aug 2011 19:57:22 +0000 (21:57 +0200)]
add isl_basic_{set,map}_partial_lex{min,max}_pw_multi_aff
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 16 Aug 2011 09:33:12 +0000 (11:33 +0200)]
add isl_pw_multi_aff_substitute
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 16 Aug 2011 09:02:36 +0000 (11:02 +0200)]
add isl_set_substitute
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 16 Aug 2011 09:00:43 +0000 (11:00 +0200)]
add isl_local_space_substitute
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 25 Aug 2011 10:15:32 +0000 (12:15 +0200)]
isl_printer_print_pw_multi_aff: partial support for C output
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 1 Sep 2011 09:58:50 +0000 (11:58 +0200)]
add isl_pw_multi_aff
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Mon, 26 Sep 2011 11:49:33 +0000 (13:49 +0200)]
add isl_pw_*_get_dim_id
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Mon, 26 Sep 2011 11:48:54 +0000 (13:48 +0200)]
add isl_pw_*_get_dim_name
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Mon, 26 Sep 2011 11:30:39 +0000 (13:30 +0200)]
add isl_pw_*_has_tuple_id
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 21 Sep 2011 09:10:37 +0000 (11:10 +0200)]
add isl_pw_*_set_dim_id
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 8 Sep 2011 12:43:42 +0000 (14:43 +0200)]
add isl_pw_*_get_tuple_id
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 25 Aug 2011 10:09:18 +0000 (12:09 +0200)]
add isl_pw_*_get_tuple_name
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 1 Sep 2011 09:59:56 +0000 (11:59 +0200)]
add isl_basic_map_from_multi_aff
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 1 Sep 2011 09:49:19 +0000 (11:49 +0200)]
add isl_multi_aff
An isl_multi_aff object represents a sequence of
zero or more affine expressions, all defined on the same domain space.
As such, they are different from isl_aff_lists as the elements in a list
may have different spaces and no space information is available at all
about empty lists.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 1 Sep 2011 09:34:57 +0000 (11:34 +0200)]
isl_pw_templ.c: rename align_params_pw_pw_and to include PW
This allows isl_pw_templ.c to be included twice from the same file.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 21 Sep 2011 14:53:11 +0000 (16:53 +0200)]
add isl_map_dim_max
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 21 Sep 2011 16:01:29 +0000 (18:01 +0200)]
privately export isl_basic_map_foreach_lexopt
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Fri, 30 Sep 2011 15:24:36 +0000 (17:24 +0200)]
add isl_basic_map_eliminate
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Fri, 30 Sep 2011 14:56:20 +0000 (16:56 +0200)]
add isl_basic_map_has_dim_id
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 2 Oct 2011 08:45:43 +0000 (10:45 +0200)]
Merge branch 'maint'
Sven Verdoolaege [Sat, 1 Oct 2011 08:29:16 +0000 (10:29 +0200)]
isl_basic_map_eliminate_vars: gauss result
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sat, 24 Sep 2011 16:08:08 +0000 (18:08 +0200)]
add isl_aff_set_dim_id
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sat, 24 Sep 2011 16:07:52 +0000 (18:07 +0200)]
add isl_local_space_set_dim_id
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sat, 24 Sep 2011 09:42:30 +0000 (11:42 +0200)]
add isl_map_find_dim_by_name
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sat, 24 Sep 2011 09:01:01 +0000 (11:01 +0200)]
add isl_space_find_dim_by_name
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 21 Sep 2011 12:15:20 +0000 (14:15 +0200)]
export isl_pw_aff_intersect_domain
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 21 Sep 2011 12:13:53 +0000 (14:13 +0200)]
export isl_pw_aff_plain_is_equal
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Mon, 26 Sep 2011 11:39:30 +0000 (13:39 +0200)]
add generic isl_pw_*_set_tuple_id
For consistency, this means we need to change the prototype of
isl_pw_aff_set_tuple_id.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 22 Sep 2011 09:53:09 +0000 (11:53 +0200)]
isl_set_intersect: properly handle intersection with parameter domains
isl_set_intersect currently allows one of its arguments to be a parameter
domain while the other is not. This will probably change in future, but
while we still support this situation, we should make sure that the result
lives in the same space as that of the argument that is not a parameter domain.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sat, 24 Sep 2011 08:54:00 +0000 (10:54 +0200)]
isl_access_info_compute_flow: align parameters of input may access relations
This part was missing from 13124d3 (isl_access_info_compute_flow: align
parameters of input access relations, Thu Aug 11 14:03:55 2011 +0200).
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Fri, 23 Sep 2011 08:53:33 +0000 (10:53 +0200)]
Merge branch 'maint'
Sven Verdoolaege [Fri, 23 Sep 2011 08:34:48 +0000 (10:34 +0200)]
fix isl_set_involves_dims
It would only consider the set to involve a dim if _all_ its subsets
would involve the dim, while it should consider the set to involve a
dim if _any_ of its subsets do.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 22 Sep 2011 14:31:29 +0000 (16:31 +0200)]
add isl_set_plain_is_fixed
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 22 Sep 2011 10:00:28 +0000 (12:00 +0200)]
add isl_set_from_params
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 22 Sep 2011 10:00:15 +0000 (12:00 +0200)]
add isl_space_set_from_params
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 22 Sep 2011 11:15:02 +0000 (13:15 +0200)]
Merge branch 'maint'
Sven Verdoolaege [Tue, 20 Sep 2011 13:59:19 +0000 (15:59 +0200)]
isl_qpolynomial_morph: properly handle denominators in morph
Before, the denominator would get ignored during the morphing of the divs.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 20 Sep 2011 13:58:11 +0000 (15:58 +0200)]
add isl_mat_diag
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 1 Sep 2011 09:42:16 +0000 (11:42 +0200)]
isl_tab_pip.c: sol_for_add: correctly set denominator
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 1 Sep 2011 09:41:22 +0000 (11:41 +0200)]
isl_tab_pip.c: sol_map_add: fix double free on error path
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 18 Sep 2011 12:41:05 +0000 (14:41 +0200)]
isl_stream: accept "==" token
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Sun, 18 Sep 2011 14:29:30 +0000 (16:29 +0200)]
add isl_union_pw_*_plain_is_equal
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Mon, 5 Sep 2011 10:26:23 +0000 (12:26 +0200)]
add isl_pw_*_plain_is_equal
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Mon, 5 Sep 2011 10:12:47 +0000 (12:12 +0200)]
add isl_set_plain_cmp
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 8 Sep 2011 14:55:41 +0000 (16:55 +0200)]
add isl_map_params
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Thu, 15 Sep 2011 12:25:51 +0000 (14:25 +0200)]
isl_stream_next_token: treat "-0" as two tokens '-' and '0'
Usually, a '-' followed by a sequence of digits is treated as a single
(negative) number. When parsing affine expressions, negative numbers
are allowed without a preceding '+'. However, if "-0" is parsed as '0'
then there is no way of distinguishing between "-0" and "0" and so "-0"
was not allowed in affine expressions.
We now treat "-0" as two tokens, so that it can be used in affine expressions.
The flip side is that "-0" can no longer be used where a single integer
is expected.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 14 Sep 2011 17:36:53 +0000 (19:36 +0200)]
drop "nparam" argument from isl_{set,map}_read_from_{file,str}
The "nparam" argument is removed because it is inconsistent with
the rest of the API.
The "nparam" argument was only ever useful for reading sets
represented in the original PolyLib format.
For reading maps, it was of limited use because there was no way
of specifying the number of input dimensions.
In both the isl format and the extended PolyLib format, the number
of parameters can be read off from the input.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 14 Sep 2011 17:50:57 +0000 (19:50 +0200)]
doc: document *_move_dims functions
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Fri, 9 Sep 2011 12:55:16 +0000 (14:55 +0200)]
drop isl_div abstraction
The isl_div abstraction has been superseded by the isl_aff abstraction,
so use that instead.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 13 Sep 2011 08:39:55 +0000 (10:39 +0200)]
isl_{in,}equality_alloc: take isl_local_space intead of isl_space
This makes it easier to construct constraints involving existentially
quantified variables.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 13 Sep 2011 10:51:12 +0000 (12:51 +0200)]
doc: update example code to change in return type of isl_constraint_set_*
The change happened way back in e2a2317 (reimplement isl_constraint in terms
of isl_aff, Fri Jul 1 17:20:52 2011 +0200).
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Fri, 9 Sep 2011 15:45:51 +0000 (17:45 +0200)]
Merge branch 'maint'
Sven Verdoolaege [Fri, 9 Sep 2011 15:40:06 +0000 (17:40 +0200)]
provide full prototypes for some functions with no arguments
Reported-by: Zbigniew Chamski <zbigniew.chamski@gmail.com>
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Fri, 2 Sep 2011 10:58:29 +0000 (12:58 +0200)]
doc: fix typo
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Wed, 7 Sep 2011 14:38:50 +0000 (16:38 +0200)]
add isl_set_gist_params
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 30 Aug 2011 14:47:59 +0000 (16:47 +0200)]
make isl_pw_* object live in a map space
Before, the space of an isl_pw_* object was that of the cells.
Conceptually, however, they map elements from these cells to something
and so their space should be that of a map.
This is especially useful for isl_pw_multi_aff objects that will
be introduced later. Since we want to be able to convert back and forth
between such objects and isl_maps, we need to keep track of the
complete map space.
isl_aff, isl_qpolynomial and isl_qpolynomial_fold objects now
also live in a map space, although internally, they still keep
track of the domain space.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 30 Aug 2011 12:26:18 +0000 (14:26 +0200)]
add isl_space_extend_domain_with_range
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 30 Aug 2011 12:25:50 +0000 (14:25 +0200)]
add isl_space_is_domain
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Tue, 30 Aug 2011 07:59:43 +0000 (09:59 +0200)]
add isl_local_space_is_set
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Mon, 29 Aug 2011 15:57:15 +0000 (17:57 +0200)]
add isl_local_space_domain
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Mon, 29 Aug 2011 13:34:59 +0000 (15:34 +0200)]
implement isl_constraint on top of isl_vec
In e2a2317 (reimplement isl_constraint in terms of isl_aff,
Fri Jul 1 17:20:52 2011 +0200), were reimplemented on top of isl_aff
because both objects share many features.
There are also some difference, however, and it is therefore better
to implement isl_constraint directly on top of isl_vec.
In particular, an isl_aff has a denominator, whereas an isl_constraint
does not. Although the denominator of an isl_aff is usually 1, it may
not always be and could therefore lead to subtle bugs.
Moreover, the space of an isl_constraint may be either that of
a set or a map, whereas, in principle, the domain of isl_aff is always
a set. We allowed the domain of an isl_aff to be a map when it was
used to implement an isl_constraint, but this was not very clean.
This would become even more dramatic when we change the space of an isl_aff
to be that of a map.
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
Sven Verdoolaege [Mon, 29 Aug 2011 13:32:39 +0000 (15:32 +0200)]
add isl_vec_neg
Signed-off-by: Sven Verdoolaege <skimo@kotnet.org>
projects / platform / upstream / isl.git / log