casereader numbering
|
View:
New views
12 Messages
—
Rating Filter:
Alert me
|
 |
casereader numbering
by John Darrington
::
Rate this Message:
I've started rewriting EXAMINE to make it less crappy.
One thing that would make it easier would be a function of the form struct casereader * append_case_numbers (struct casereader *cr); which returns a new casereader which is identical to CR except that it has one extra column which contains the ordinal number of each case. For example if CR contains x y z . a b c d the return will contain x y 1 z . 2 a b 3 c d 4 Such a function would also simply the implementation of RANK. But it's not obvious to me how to create such a function. Is this a feasible thing to do? J' -- PGP Public key ID: 1024D/2DE827B3 fingerprint = 8797 A26D 0854 2EAB 0285 A290 8A67 719C 2DE8 27B3 See http://pgp.mit.edu or any PGP keyserver for public key. _______________________________________________ pspp-dev mailing list pspp-dev@... http://lists.gnu.org/mailman/listinfo/pspp-dev |
|
|
Re: casereader numbering
by Jason Stover-3
::
Rate this Message:
|
|
|
Re: casereader numbering
by Ben Pfaff
::
Rate this Message:
|
|
|
Re: casereader numbering
by John Darrington
::
Rate this Message:
On Thu, Jul 17, 2008 at 03:18:33PM -0400, Jason Stover wrote:
This would be valuable for other procedures, too. Anything that could use permutations could make use of such a function. Also, it would be useful for bootstrap and jacknife tests. How would numbering the cases will help for those applications? I would have thought that some kind of random/low discrepency sequence iterator would be required. J' -- PGP Public key ID: 1024D/2DE827B3 fingerprint = 8797 A26D 0854 2EAB 0285 A290 8A67 719C 2DE8 27B3 See http://pgp.mit.edu or any PGP keyserver for public key. _______________________________________________ pspp-dev mailing list pspp-dev@... http://lists.gnu.org/mailman/listinfo/pspp-dev |
|
|
Re: casereader numbering
by Jason Stover-3
::
Rate this Message:
|
|
|
Re: casereader numbering
by Ben Pfaff
::
Rate this Message:
|
|
|
Re: casereader numbering
by John Darrington
::
Rate this Message:
On Tue, Jul 22, 2008 at 10:13:31PM -0700, Ben Pfaff wrote:
John Darrington <john@...> writes: > One thing that would make it easier would be a function of the form > > struct casereader * append_case_numbers (struct casereader *cr); > > which returns a new casereader which is identical to CR except that it > has one extra column which contains the ordinal number of each case. Here is a patch to try. I haven't had a chance to test it, but I think that it should work. If it works for you I'll push it to the repository. On initial tests, it appears to work fine, except that I would have expected casereader_get_value_cnt on the new casereader to return 1 more than that of the old one. But that's not what I am experiencing. J' -- PGP Public key ID: 1024D/2DE827B3 fingerprint = 8797 A26D 0854 2EAB 0285 A290 8A67 719C 2DE8 27B3 See http://pgp.mit.edu or any PGP keyserver for public key. _______________________________________________ pspp-dev mailing list pspp-dev@... http://lists.gnu.org/mailman/listinfo/pspp-dev |
|
|
Re: casereader numbering
by Ben Pfaff
::
Rate this Message:
|
|
|
Re: casereader numbering
by John Darrington
::
Rate this Message:
On Wed, Jul 23, 2008 at 07:24:21AM -0700, Ben Pfaff wrote:
> On initial tests, it appears to work fine, except that I would have > expected casereader_get_value_cnt on the new casereader to return 1 more > than that of the old one. But that's not what I am experiencing. Er, I would expect that too. On inspection, the code looks correct; I don't see how casereader_get_value_cnt() could return value different from that. Huh. Maybe I made a mistake then. I'll have closer look later. J' -- PGP Public key ID: 1024D/2DE827B3 fingerprint = 8797 A26D 0854 2EAB 0285 A290 8A67 719C 2DE8 27B3 See http://pgp.mit.edu or any PGP keyserver for public key. _______________________________________________ pspp-dev mailing list pspp-dev@... http://lists.gnu.org/mailman/listinfo/pspp-dev |
|
|
Re: casereader numbering
by John Darrington
::
Rate this Message:
On Thu, Jul 24, 2008 at 07:06:05AM +0800, John Darrington wrote:
On Wed, Jul 23, 2008 at 07:24:21AM -0700, Ben Pfaff wrote: > On initial tests, it appears to work fine, except that I would have > expected casereader_get_value_cnt on the new casereader to return 1 more > than that of the old one. But that's not what I am experiencing. Er, I would expect that too. On inspection, the code looks correct; I don't see how casereader_get_value_cnt() could return value different from that. Huh. Maybe I made a mistake then. I'll have closer look later. On closer inspection it turns out that the culprit is the function sort_execute : struct casereader * sort_execute (struct casereader *input, struct case_ordering *ordering); Instead of returning a casereader the same width as INPUT, it returns one with the width associated with ORDERING. I don't see any valid reason to for a case_ordering to be aware of the value_cnt, so I'm proposing this patch which seems to fix this problem. diff --git a/src/data/case-ordering.c b/src/data/case-ordering.c index 7b3948c..c4a716e 100644 --- a/src/data/case-ordering.c +++ b/src/data/case-ordering.c @@ -37,8 +37,6 @@ struct sort_key /* A set of criteria for ordering cases. */ struct case_ordering { - size_t value_cnt; /* Number of `union value's per case. */ - /* Sort keys. */ struct sort_key *keys; size_t key_cnt; @@ -49,10 +47,9 @@ struct case_ordering contains no variables, so that all cases will compare as equal. */ struct case_ordering * -case_ordering_create (const struct dictionary *dict) +case_ordering_create (void) { struct case_ordering *co = xmalloc (sizeof *co); - co->value_cnt = dict_get_next_value_idx (dict); co->keys = NULL; co->key_cnt = 0; return co; @@ -63,7 +60,6 @@ struct case_ordering * case_ordering_clone (const struct case_ordering *orig) { struct case_ordering *co = xmalloc (sizeof *co); - co->value_cnt = orig->value_cnt; co->keys = xmemdup (orig->keys, orig->key_cnt * sizeof *orig->keys); co->key_cnt = orig->key_cnt; return co; @@ -80,15 +76,6 @@ case_ordering_destroy (struct case_ordering *co) } } -/* Returns the number of `union value's in the cases that case - ordering CO compares (taken from the dictionary used to - construct it). */ -size_t -case_ordering_get_value_cnt (const struct case_ordering *co) -{ - return co->value_cnt; -} - /* Compares cases A and B given case ordering CO and returns a strcmp()-type result. */ int diff --git a/src/data/case-ordering.h b/src/data/case-ordering.h index 026cd89..f49f265 100644 --- a/src/data/case-ordering.h +++ b/src/data/case-ordering.h @@ -32,7 +32,7 @@ enum sort_direction }; /* Creation and destruction. */ -struct case_ordering *case_ordering_create (const struct dictionary *); +struct case_ordering *case_ordering_create (void); struct case_ordering *case_ordering_clone (const struct case_ordering *); void case_ordering_destroy (struct case_ordering *); diff --git a/src/language/stats/rank.q b/src/language/stats/rank.q index 5bc88c4..cb63949 100644 --- a/src/language/stats/rank.q +++ b/src/language/stats/rank.q @@ -261,7 +261,7 @@ rank_cmd (struct dataset *ds, const struct case_ordering *sc, /* Sort this split group by the BY variables as primary keys and the rank variable as secondary key. */ - ordering = case_ordering_create (d); + ordering = case_ordering_create (); for (j = 0; j < n_group_vars; j++) case_ordering_add_var (ordering, group_vars[j], SRT_ASCEND); case_ordering_add_var (ordering, @@ -778,7 +778,7 @@ cmd_rank (struct lexer *lexer, struct dataset *ds) /* Put the active file back in its original order. Delete our sort key, which we don't need anymore. */ { - struct case_ordering *ordering = case_ordering_create (dataset_dict (ds)); + struct case_ordering *ordering = case_ordering_create (); struct casereader *sorted; case_ordering_add_var (ordering, order, SRT_ASCEND); /* FIXME: loses error conditions. */ diff --git a/src/language/stats/sort-criteria.c b/src/language/stats/sort-criteria.c index c84f71d..fd8c7c5 100644 --- a/src/language/stats/sort-criteria.c +++ b/src/language/stats/sort-criteria.c @@ -39,7 +39,7 @@ struct case_ordering * parse_case_ordering (struct lexer *lexer, const struct dictionary *dict, bool *saw_direction) { - struct case_ordering *ordering = case_ordering_create (dict); + struct case_ordering *ordering = case_ordering_create (); const struct variable **vars = NULL; size_t var_cnt = 0; diff --git a/src/math/merge.c b/src/math/merge.c index d56a78c..4fc7c8d 100644 --- a/src/math/merge.c +++ b/src/math/merge.c @@ -44,16 +44,18 @@ struct merge struct case_ordering *ordering; struct merge_input inputs[MAX_MERGE_ORDER]; size_t input_cnt; + size_t value_cnt; }; static void do_merge (struct merge *m); struct merge * -merge_create (const struct case_ordering *ordering) +merge_create (const struct case_ordering *ordering, size_t value_cnt) { struct merge *m = xmalloc (sizeof *m); m->ordering = case_ordering_clone (ordering); m->input_cnt = 0; + m->value_cnt = value_cnt; return m; } @@ -95,8 +97,7 @@ merge_make_reader (struct merge *m) } else if (m->input_cnt == 0) { - size_t value_cnt = case_ordering_get_value_cnt (m->ordering); - struct casewriter *writer = mem_writer_create (value_cnt); + struct casewriter *writer = mem_writer_create (m->value_cnt); r = casewriter_make_reader (writer); } else @@ -129,7 +130,7 @@ do_merge (struct merge *m) assert (m->input_cnt > 1); - w = tmpfile_writer_create (case_ordering_get_value_cnt (m->ordering)); + w = tmpfile_writer_create (m->value_cnt); for (i = 0; i < m->input_cnt; i++) taint_propagate (casereader_get_taint (m->inputs[i].reader), casewriter_get_taint (w)); diff --git a/src/math/merge.h b/src/math/merge.h index c9c9c48..18322e8 100644 --- a/src/math/merge.h +++ b/src/math/merge.h @@ -18,11 +18,12 @@ #define MATH_MERGE_H 1 #include <stdbool.h> +#include <stddef.h> struct case_ordering; struct casereader; -struct merge *merge_create (const struct case_ordering *); +struct merge *merge_create (const struct case_ordering *, size_t); void merge_destroy (struct merge *); void merge_append (struct merge *, struct casereader *); struct casereader *merge_make_reader (struct merge *); diff --git a/src/math/sort.c b/src/math/sort.c index e03ef57..10b8a12 100644 --- a/src/math/sort.c +++ b/src/math/sort.c @@ -41,6 +41,7 @@ int max_buffers = INT_MAX; struct sort_writer { + size_t value_cnt; struct case_ordering *ordering; struct merge *merge; struct pqueue *pqueue; @@ -52,7 +53,7 @@ struct sort_writer static struct casewriter_class sort_casewriter_class; -static struct pqueue *pqueue_create (const struct case_ordering *); +static struct pqueue *pqueue_create (const struct case_ordering *, size_t); static void pqueue_destroy (struct pqueue *); static bool pqueue_is_full (const struct pqueue *); static bool pqueue_is_empty (const struct pqueue *); @@ -62,15 +63,15 @@ static void pqueue_pop (struct pqueue *, struct ccase *, casenumber *); static void output_record (struct sort_writer *); struct casewriter * -sort_create_writer (struct case_ordering *ordering) +sort_create_writer (struct case_ordering *ordering, size_t value_cnt) { - size_t value_cnt = case_ordering_get_value_cnt (ordering); struct sort_writer *sort; sort = xmalloc (sizeof *sort); + sort->value_cnt = value_cnt; sort->ordering = case_ordering_clone (ordering); - sort->merge = merge_create (ordering); - sort->pqueue = pqueue_create (ordering); + sort->merge = merge_create (ordering, value_cnt); + sort->pqueue = pqueue_create (ordering, value_cnt); sort->run = NULL; sort->run_id = 0; case_nullify (&sort->run_end); @@ -118,8 +119,7 @@ sort_casewriter_convert_to_reader (struct casewriter *writer, void *sort_) if (sort->run == NULL && sort->run_id == 0) { /* In-core sort. */ - sort->run = mem_writer_create (case_ordering_get_value_cnt ( - sort->ordering)); + sort->run = mem_writer_create (casewriter_get_value_cnt (writer)); sort->run_id = 1; } while (!pqueue_is_empty (sort->pqueue)) @@ -151,8 +151,7 @@ output_record (struct sort_writer *sort) } if (sort->run == NULL) { - sort->run = tmpfile_writer_create (case_ordering_get_value_cnt ( - sort->ordering)); + sort->run = tmpfile_writer_create (sort->value_cnt); sort->run_id = min_run_id; } @@ -176,7 +175,8 @@ static struct casewriter_class sort_casewriter_class = struct casereader * sort_execute (struct casereader *input, struct case_ordering *ordering) { - struct casewriter *output = sort_create_writer (ordering); + struct casewriter *output = + sort_create_writer (ordering, casereader_get_value_cnt (input)); casereader_transfer (input, output); return casewriter_make_reader (output); } @@ -201,14 +201,14 @@ static int compare_pqueue_records_minheap (const void *a, const void *b, const void *pq_); static struct pqueue * -pqueue_create (const struct case_ordering *ordering) +pqueue_create (const struct case_ordering *ordering, size_t value_cnt) { struct pqueue *pq; pq = xmalloc (sizeof *pq); pq->ordering = case_ordering_clone (ordering); pq->record_cap - = settings_get_workspace_cases (case_ordering_get_value_cnt (ordering)); + = settings_get_workspace_cases (value_cnt); if (pq->record_cap > max_buffers) pq->record_cap = max_buffers; else if (pq->record_cap < min_buffers) diff --git a/src/math/sort.h b/src/math/sort.h index 7f7b2f8..ea2c16b 100644 --- a/src/math/sort.h +++ b/src/math/sort.h @@ -25,7 +25,7 @@ struct case_ordering; extern int min_buffers ; extern int max_buffers ; -struct casewriter *sort_create_writer (struct case_ordering *); +struct casewriter *sort_create_writer (struct case_ordering *, size_t value_cnt); struct casereader *sort_execute (struct casereader *, struct case_ordering *); #endif /* math/sort.h */ -- PGP Public key ID: 1024D/2DE827B3 fingerprint = 8797 A26D 0854 2EAB 0285 A290 8A67 719C 2DE8 27B3 See http://pgp.mit.edu or any PGP keyserver for public key. _______________________________________________ pspp-dev mailing list pspp-dev@... http://lists.gnu.org/mailman/listinfo/pspp-dev |
|
|
Re: casereader numbering
by Ben Pfaff
::
Rate this Message:
|
|
|
Re: casereader numbering
by Ben Pfaff
::
Rate this Message:
|
signature.asc (196 bytes) 
| Free Forum Powered by Nabble | Forum Help |