Philipp Rotmann, Linksystem Muenchen wrote:
> Hi there,
>
> it's me again ...
>
> On 19.01.01 (18:08), David Guerizec wrote:
<snip>
>
>> The next step is to extract/export the SG0 from the old database
>> (containing your clients sites) into an xml.gz file.
>
>
> All done w/o problems now.
good ;)
One thing though, I forgot to tell you to dump your DB in SQL format at
this stage, just in case something goes wrong at the next step
(importing with repligard), you could then start from that point with a
clean database instead of going thru all the init process again...
>
>> You import that xml.gz file into the new database and into the
>> new SG you just created (use the username and password of the
>> adminuser in the repligard.conf file), and ...
>
>
> .... and repligard coredumps again:
>
> dilbert:~/xsg > repligard -c ../sgmove-repligard.conf -i SG0.xml.gz
> Message: Reading config file ../sgmove-repligard.conf
> Message: Importing SG0.xml.gz
>
> REPLIGARD/READ-ERROR **: not well-formed (invalid token) in line 50366
> aborting...
> Aborted (core dumped)
>
> Line 50366 contains
>
> <author>1267692cc7fa9f13507590f46c5e497e</author>
>
> Could there be any problem with the sheer size of the repligard
> XML file? The same repligard binary created it by exporting my
> existing Midgard database, and now it raises an error because of
> 'well-formedness'???
I already encountered this kind of problem, and it was in fact the 'url'
field of an article that contained an '&' character...
(BTW, Alexander, this 'url' field should be defined as CDATA cause some
people use it as an 'extra' field...)
Can you check around this line 50366 to see if a non-CDATA field
contains a non-alphanumeric character ?
>
>
> phr
>
--
Best Regards,
David Guerizec Open Source Developer
Midgard core developer http://www.midgard-project.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
