crashes on sorting or other action requiring downloading dat
Brought to you by:
louburnard
I have indexed the Dictionary of Old English for data retrieval in Xaira (ca 4,5 mil words, over 3000 files). Search works fine, but when I try sorting by collumns or creating a partition, Xaira reports "downloading" for a few seconds and crashes.
"Microsoft Visual C++ Runtime Error" in xaira.exe, sometimes you can also get two memory locations (?) 0x40000015 and 0x7857bea4.
Thanks!
Please supply more information about the software version and platform you are using. It sounds like a client side memory leak: if you build and index a smaller version of the corpus (with just say 100 files) do you see the same behaviour?
Thanks for the reply! I use Xaira client 1.25 (the same behaviour reproduce also with ver. 1.23) on Windows XP SP3. If I use the same data with a small number of files, everything works fine (in case of a large number of files and completely different files, the crash happens as expected). It seems there is indeed a memory leak while Xaira reports "Downloading" on its status bar. Otherwise the corpora are fully searchable.
OK, so this is a memory handling issue on the windows client. Trouble is that we're not doing any more development on the windows client, so I don't know how quickly we can resolve it. Can you do anything to increase the amount of memory available to your client machine?
The client machine has about 4gb of RAM, so unless it is some problem caused by memory allocation in the OS, I don't think the total amount of memory available could be the issue.
Is there a newer version available for a different platform? I'd be happy to use a linux version, although our main reason for using Xaira was its accessibility (GUI & Windows) that allows us to teach it even to our students without much IT experience.
As I said in an earlier reply, we don't see this problem when accessing the BNC via the Windows client. In that case, the server is running on a linux box accessed over the network, so I think the problem has to be somewhere in the standalone server/client set up. Building the server on Linux is not too hard, once you have all the right bits and pieces, but you have to build it from source as we don't have a binary. If that works, your students could then go on accessing the corpus using the Windows interface. You shouldn't need to reindex the corpus even, since the indexes are (supposed to be) platform independent.
Thanks for the suggestion, I'll see if we can manage to get the server installed on an accessible Linux server (this poses some problems with working from home, but that is our problem). However, one thing baffles me - if this is a server problem, isn't it surprising, that it crashes the client? Or is in the Windows environment the client in the same executable as the server?
Just in case there are other Windows users/loosers like me out there who run into similar problems. This is a weird error, but not insurmountable. Time to process data seems to be an issue. After testing this on several systems and with different parameters I conclude that slow (USB, networked, etc) drives cause problems, more columns displayed also cause problems and newer versions cause problems. Also, if you are unable to create paritions on your system even with just the two basic columns, create it elsewhere (seems to be rock solid on Win2k), transfer the partition files on your system and as long as you don't sort by columns, you can use the partitions without any problems.
outside openings delivering employment #file=C:\baobao\keyword.txt of your Closed circuit cheap beats by dre http://www.sunnertech.com/about.asp