I have data that is using a different charset.
When displaying this data in a simple table, Splunk parses it as string and applies same rules when sorting it, by bytes.
This causes the text to be sorted like this:
A, B, C .... X, Y, Z, local_char_1, local_char_2 etc.
That's correct when just checking bytes (unicode), however our alphabet is different. Is it possible to modify the sort command's behavior so it sorts text like this:
A, B, C, local_char_1, D, E ....
The issue is that I want to allow the user to sort as he/she wants in the interface, by any column that is displayed in the table. I can maybe use a hidden variable that is populated by my external script but that will work for first display only, as far as I understand.
To use a fancy word, you're wanting to change the "collating sequence" for sort. As far as I know this is not possible directly out of the box. This type of internationalization support could make a good enhancement request. There's a chance that the collating sequence is governed by the language settings in Splunk, but I've not tested.
In a worst case, you could implement your own sort algorithm using a custom search command. Then you are free to define whatever collating sequence you require.
Thanks - does that mean I just create a new command and use it like this:
old search | mysearch
How will this work when the user clicks on the sort button in the table? Generally I have problems only with text fields (due to the character set, obviously), with numbers it (of course) works ok.
Can you use the fields command after table to mandate the order of the fields?