Hi,

After reviewing all the other alternatives like SQL and Python UDF, I was either stuck on performance with SQL UDF or on usability with Python UDF (unable to use with aggregation, and not such great performance with dates), 

so I decided to go the hard way with C functions, as a bonus it will give me the possibility to change the functionalities without worrying about dependencies, which was not the case in other languages.

The purpose is to create a set of formatting functions for Year, Quarter, Month, Week and Day brackets, and of course i need to create the bulk version of each function for performance. 

Starting from the MTIMEdate_extract_year_bulk, now i have the simple function working, and successfully calling it from mclient:

str

UDFyearbracket(str *ret, const date *v)

{

if (*v == date_nil) {

*ret = GDKstrdup(str_nil);

} else {

int year;

fromdate(*v, NULL, NULL, &year);

*ret = (str) GDKmalloc(15);

sprintf(*ret, "%d", year);

}

return MAL_SUCCEED;

}

For the bulk version i get an error in the log: gdk_atoms.c:1345: strPut: Assertion `(v[i] & 0x80) == 0' failed.

str

UDFBATyearbracket(bat *ret, const bat *bid)

{

BAT *b, *bn;

BUN i,n;

str *y;

const date *t;

if ((b = BATdescriptor(*bid)) == NULL)

throw(MAL, "UDF.BATyearbracket", "Cannot access descriptor");

n = BATcount(b);

bn = COLnew(b->hseqbase, TYPE_str, BATcount(b), TRANSIENT);

if (bn == NULL) {

BBPunfix(b->batCacheid);

throw(MAL, "UDF.BATyearbracket", "memory allocation failure");

}

bn->tnonil = 1;

bn->tnil = 0;

t = (const date *) Tloc(b, 0);

y = (str *) Tloc(bn, 0);

for (i = 0; i < n; i++) {

if (*t == date_nil) {

*y = GDKstrdup(str_nil);

} else

UDFyearbracket(y, t);

if (strcmp(*y, str_nil) == 0) {

bn->tnonil = 0;

bn->tnil = 1;

}

y++;

t++;

}

BATsetcount(bn, (BUN) (y - (str *) Tloc(bn, 0)));

bn->tsorted = BATcount(bn)<2;

bn->trevsorted = BATcount(bn)<2;

BBPkeepref(*ret = bn->batCacheid);

BBPunfix(b->batCacheid);

return MAL_SUCCEED;

}

PS: I am not a c expert but i can find my way with basic operations and pointers.

Any help or suggestions is appreciated.

Thank you.

_______________________________________________
users-list mailing list
users-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/users-list