>>>>>>>>>>>>>>>>>> Original Message <<<<<<<<<<<<<<<<<<
On 1/18/01, 9:20:15 PM, Bhavin Turakhia <[EMAIL PROTECTED]> wrote
regarding RE: [midgard-user] Cannot create Attachments: Internal error:
> PLEASE CHECK AT BOTTOM FOR THE LATEST UPDATE - i tried what you said
Armand
> but it still does not work. Chck below
> For all the above additions the Mysql entries work percfectly. ie there
is
> no problem in MySQL. For instance for every host i have added or
attachment
> i have added there have been entries made in the hosts and the blobs
table.
> I checked that through the mysql client.
> The version i am running is 3.22.32 which i think is ok.
> As per your suggestions i also started logging and did a tail -f and then
> went ahead and added an attachment and got this -
> 398 Query INSERT INTO blobs (ptable,pid,name,title,loca
> tion,mimetype,author,created,score, sitegroup) VALUES
> ('page',36,'uio','oiu','3/
> 7/7c757c8b0700726e09e62af93d2f710a','image/gif',1,Now(),0, 0)
> 010118 12:49:10 398 Query INSERT INTO repligard
> (realm,guid,id,changed,
> action) VALUES
> ('blobs','d2d81167a2c29964de8226622e9ba99e',1082438592,NULL,'crea
> te')
> Both these queries went in successfully as i checked the database and
these
> rows have been inserted.
> So I am back to square 1.
Hmm.. I'm at a loss here. The 'Internal error' and the attachments not
being created are certainly connected. I was guessing that mysql somehow
performs the queries successfully, but still returns an error according
to midgard-lib. It makes sense not to save the actual blob if it looks
like the blob-record could not be created.
So the question boils down to why you get the internal error. I've
searched the source for midgard-php, which is where the error is
generated from. I've only found one instance of
RETURN_FALSE_BECAUSE(MGD_ERR_INTERNAL) in the code (mgd_main.c) that
pertains to a create statement (lot's of disclaimers, I'm not too good at
reading c):
void php_midgard_create(pval * return_value, const char *table,
const char *fields, const char *values, ...)
{
va_list args;
int id;
CHECK_MGD;
va_start(args, values);
id = mgd_vcreate(MGD_RCFG->mgd, table, fields, values, args);
va_end(args);
if (!id) {
RETURN_FALSE_BECAUSE(MGD_ERR_INTERNAL);
ap_log_rerror(APLOG_MARK, APLOG_DEBUG|APLOG_NOERRNO,
GLOBAL(php3_rqst),
"Midgard: create of %s failed",
table ? table : "<null class>");
}
RETURN_LONG(id);
}
so mgd_vcreate does not return the id of the created record. mgd_vcreate
lives in midgard.c (from midgard-lib)
int mgd_vcreate(midgard * mgd, const char *table,
const char *fields, const char *values, va_list args)
{
const char *command = NULL;
int rv;
midgard_pool *pool;
pool = mgd_alloc_pool();
if (!pool)
return 0;
/* format command */
#if HAVE_MIDGARD_SITEGROUPS
if (strcmp(table, "sitegroup") == 0) {
if (mgd->root)
command = mgd_format(mgd, pool,
"INSERT INTO $s ($s) VALUES ($s)",
table, fields, values);
}
else
command = mgd_format(mgd, pool,
"INSERT INTO $s ($s, sitegroup) VALUES ($s, $d)",
table, fields, values, mgd->sitegroup);
#else
command = mgd_format(mgd, pool,
"INSERT INTO $s ($s) VALUES ($s)",
table, fields, values);
#endif
if (!command) {
mgd_free_pool(pool);
return 0;
}
/* execute command */
rv = mgd_vexec(mgd, command, args);
mgd_free_pool(pool);
return rv ? mysql_insert_id(mgd->mysql) : 0;
}
So I'm still at your mysql installation not returning the expected return
value. I haven't seen this before, and I would look at stray libraries
(older version of libmysqlclient or some such). Beyond that I am afraid I
can't be of any more help. CC-ing this one to the development list,
hoping that someone will confirm/correct my findings.
Armand.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
