Spark 2.6.3, OpenFire 3.7.0. Overall impression is bad (earlier versions too). Spark almost hangs when process history data. Processing is very slow. Trying to resize history window may lead to hang this window. In addition, there is “black hole” between “current” and “conventional” history. If I disable “showing previous conversation in chat”, then Spark simply omit “current” history. When I enable “showing previous conversation in chat”, then Spark show some (but not all!) “current” history, “History” button show “old” history, and some “current” history isn’t visible at all. BTW conversation history is not too large - for example, conversation file size with example user is about 3M. I think that history in Spark needs to be improved.
Please review the forum about this. The issue is caused by a standard Java function that is slow. The only way to get around this is to move the function view history out of Spark. By teh way, why do you need to read 3 MB of historical chat ?
I am using Spark for 5 years, my impression is very good.
Spark uses XML to store the history of conversations, leading to large heavy reading files. In case of Spark crash history files can became malformed. If conversation history view cannot be improved, then better to store plain text files (timestamp / message), or use another engine, i.e. SQL DBMS. Compared to ICQ Spark’s history view implementation is almost unusable at present. Why 3MB? Believe me, 3MB, and even in XML-format - it is very little useful information. And I cannot use it because after some history size viewing is simply unusable. 3MB, or 300MB - it’s unusable. I’m using ICQ 2002a for about 10 years and it’s history implementation is simple and useful - “utile dulci”.
3 MB is a lot. It’s a known thing, that Spark works slowly with a large history. But so far no solution. It doesn’t bother me very much as i don’t have to read old history too often. But recently Spark started to stall for a second when sending/receiving messages with a contact which i have rather big conversation history with.
The problem with not seeing the whole history may be related to the limit of 5000 messages, which was set by Wolf to prevent Spark hanging. But it omits the recent messages instead of the last ones. SPARK-1407
You can disable this limit by adding maximumHistory=xxx into spark.properties and setting it to some really big number.
Jaroslav Mazurak wrote:
If conversation history view cannot be improved, then better to store plain text files (timestamp / message), or use another engine, i.e. SQL DBMS
This would be great. Submit a patch. And it would be great if it would import all the current xml conversations into a new format storage.