all of a sudden parsing of atom feeds from blogpsot/blogger fails.

I'm aggregating a couple of hundred of blogspot feeds (check http://www.manalaa.net/aggregator2/opml) and things where running smoothly until somtime in the past two days.

I reverted back to an older checkout of aggregator2 that I know works and the problem prevails so this might be caused by a change in blogspot, can anyone confirm this?

Comments

ahwayakchih’s picture

ahwayakchih commented

i just tested this one: http://1pissedarab.blogspot.com/atom.xml
and it parsed fine (created 24 items). uning 4.7 but nothing changed in parsing code between 4.6.x and 4.7.

alaa’s picture

alaa commented

funky, I tried resubmitting that feed and got

Failed to parse RSS feed 1pissedarab.blogspot.com: no element found at line 48.

as error, any ideas on what could be the problem on my side, or what do I need to check?

I'm using 4.6.3 and latest aggregator2 cvs.

alaa’s picture

alaa commented
Title: parsing blogpsot atom feeds fails » sorry my bad

sorry for wasting your time, turned out the problem was on my server, for some reason the network interface was set to 10Mb instead of 100Mb and downloading the feeds was too slow and apparently timed out with only a portion of the XML downloaded.

is there a way for aggregator2 to detect incomplete downloads and report them as such instead of trying to parse the xml portion?

ahwayakchih’s picture

ahwayakchih commented
Status: Active » Closed (fixed)

Aggregator uses expat library for parsing xml. If somethings is wrong it just logs error and does nothing more (doesn't try to add any nodes, update them or anything).
If we wanted to react on different errors, etc.. we would have to either write own parser, or add more handling for different errors. In any case that would most probably slow down process very much, and thus would make agg2 much less usable.