Re: [Social-discuss] Upgrading from Statusnet to GNU Social - when can users work with it?

[Michael Vogel]

Hi!

Am 19.07.2017 17:31, schrieb Mikael Nordfeldth:
> On 2017-07-19 11:05, Michael Vogel wrote:
> > | Shortening url entry for file id: 259624 [
> > | Deleting duplicate entries of too long URL on file id: 259624 [....
> >
> > [...]
> >
> > I have the feeling that there is some bug in the deletion of 
> > duplicated
> > entries, can you confirm this?
>
> That doesn't sound right, unless all the deleted entries are duplicates
> up to 191 characters. Some URLs are actually really long and more URLs
> to the same page may very well exceed 191 characters.
...
> So one question: Are you sure it's _every other_ File entry and not 
> just
> very similar once in a large quantity?

I'm totally sure. It deleted the first 200k entries until I stopped the 
script. I checked the content of the deleted entries and there were no 
similarities.

> If so, it surely is a bug. Maybe the output should show which URLs are
> being deleted so debugging this would be easier remotely.

I executed the query by hand that showed me the 4 long file entries with 
duplicates. Then I simply deleted the duplicate entries, then it skipped 
that part after I executed the upgrade script again.

The problem has to be the query that fetches all entries that are 
duplicates - although this query looks fine to me.

Michael