> Nicolas Desir wrote:
>
>> Hello,
>>
>> I am running a debian potato with
>> libc6 2.1.3-17, php 4.0.3, apache 1.3.9, mysql 3.22.32
>> midgard 1.4.1
>>
>> The hostname of the server was defined correctly in
>> midgard/data/configure, but I had to change it by hand in
>> the host table from localhost to the real hostanme of the
>> server. An then midgard started to work.
>
> Yes, the data packages uses 'localhost' by default.
>
>> I had a look at dbinstall, and I discover it uses User
>> Variables that seem to have been implemented only in
>> mysql 3.23.6 (mysql changelog: Added patch by Sasha for
>> user-defined variables.) and I have MySQL 3.22.32 as
>> told in the min requirements on the midgard website.
>
> I have just now installed the data package on 3.22.30. What
> user-defined variables would those be?
He says about saving result of query into intermediate variable with @
prefix and use it in the next query.
>
>> But the real troubles come whith repligard when I try
>> to import asgard or nadmin. When dbinstall try to import
>> asgard I have:
>> REPLIGARD/READ-CRITICAL **: Failed to allocate delayed resource
>> 1708eb991e646f7be949f5c5df577425 as a link pid for blobs
>
> Alexander? What could be the cause for such an message (and do you
> think repligard itself could give more details)?
I think it is already fixed in CVS.
>
>> Looking deeper inside (strace and and sources) I found that
>> repligard is doing an invalid sql request in both case
>>
>> asgard:
>> INSERT INTO page () VALUES ()
>> You have an error in your SQL syntax near \') VALUES ()\' at lin e 1
>>
>> nadmin:
>> INSERT INTO person () VALUES ()
>> You have an error in your SQL syntax near \') VALUES ()\' at lin e 1"
>>
>> Seems that the request was generated by mgd_vcreate
>> (in lib/src/midgard.c), which was called by mgd_create
>> which was called by repligard_create_fake_resource
>> (in lib/repligard/import.c) which was called by
>> rpd_read_elt_end (in read.c) that finally display the error
>> message "REPLIGARD/READ-CRITICAL ... "
>
> But this problem should be pretty easy to guard for, right Alexander?
> Under what circumstances would these fields be empty, anyway?
It was typo introduced by Ming's patch at the beginning of April, I've fixed
it in CVS several days ago.
--
Sincerely yours, Alexander Bokovoy
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
