EVE Technology Lab

Hi Shin,

First, thanks a million for picking up this project :) Big kudos :)

Secondly, what's the best way to upgrade from 1.3.0 r179?
The 1.3.1 r5 package info do not carry the upgrade path.
Is the one from 1.3.0 applicable or does it need to be customized?

The one which was included in r179 looks like this (had to sanitize xml tags into || and |- respectively)

I'm experiencing some issues as of today. If a user tries to register, or an existing user tries to update/add API to profile, or the cron runs (whichever happens first), the characters are coming back without corp tags and therefore not qualifying for any groups. Users are being removed from their groups as well....

For instance, The users are listed in the back end like: [] Your_Name (closed brackets where the corp tag would be)

Any ideas?

* Edit: Been working fine for months and nothing has changed/been updated recently. The only thing that's changed is the new eve patch but I can't imagine that would have anything to do with it...

* Edit #2: Confirmed the same problem happening on another site. Exact same issue.

Hi friends, after patch my temark is crash down, he cant fetch char list.

registration -> input keyID -> input vCode -> push "get Characters" = error()

if update forum profile we can see first string "http://clip2net.com/s/716ADm"

please help for fix it.

Well if it only started happening Today then it can't be the patch as that was before that :)

What version are you guys running ?

The XML from CCP's servers looks just the same as it did before and while I hate to say it ... it works fine for me :) - I've just tried registering a new user and that is fine too.

Can you guys clear out your tea api cache table and also run the db_clean_up script to remove any orphaned api's/toons then retry things.

Dark Angel66 ... where is that screenshot taken ... I can't see enough of the page ... I think it is just the profile page of a user ... in which case just how is there 4 toons listed for a single api ? This is why I'm suggesting the db_clean_up script.

This is the problem that forum's that use TEA and anything else that uses its process are affected by.

With TEA, the software goes to /account/Characters.xml.aspx to get all the info it needs, including corporationID, but CCP have recently changed that page so it removes some of the columns, but not the data those columns have.

So when you pull the xml, you get this...



See what happened there?

The columns ended at corporationID, but the data for allianceID, allianceName, factionID and factionName is still there.

So now what happens, is TEA and other apps goto pull /corp/CorporationSheet.xml.aspx, which includes the data "corporationID=732773091" allianceID="1220174199" allianceName="Fatal Ascension" factionID="0" factionName=" ", causing the api server to have a hissy fit.

Getting the code to ignore everything after corporationID is easy, but will likely break if/when CCP fix it.

I‘m running v1.3.1r4

If there were no registrations and API checks since patch day, I think it still is possible. I think Demitrios may have identified the issue. Doesn‘t explain though why it works in some versions of the mod, while it‘s broken in others.

I'm running an older version of TEA and made the following change to fix the issue with the changed attributes

In file 'Sources/TEAC.php' I changed line 573 and 1006 from

$char = explode('" />', $char[1], 2);

to

$char = explode('" allianceID="', $char[1], 2);

This seems to have fixed the problem for now but the real fix would be to properly process XML as XML rather than rely on the ordering of CCP's API output.

I haven't looked at the updated version of the code by the OP of this thread. I assume it has the same problem from the comments above this.

Ah Ha ... nice find Demitrios. I looked too late in the process sequence ... I was rushing to take a look before heading to the Opticians and didn't track back far enough.

I suppose you could add in some code to truncate after the first space and use that for corpID but as you say we shouldn't have to coz it is just broken xml.



Oreamnos Amric - the new versions do exactly what you say ... they put the returned XML in a SimpleXML object and use that for the parsing ... much nicer than the older way of doing it but it means that if the XML is invalid there are problems which is exactly what we are seeing here.

Treating XML as XML you say? Well I'm sold

There isn't a massive amount you can do with malformed XML except behave properly. I'd seen the old TEA remove everyone's access when the API didn't return any information at all. Probably due to the same code as was hit here. With a proper XML parser you at least know if you have received junk data and can behave properly (i.e. do nothing rather than mess up everyone's access).

I'm assuming we just basically need to wait for CCP to fix this? If we change things now it's just going to break again when they change/fix things on their end, right?