Recommended way of working with a lot of Waveform attributes

[dfousek]

Hi !

I want to include in my 1D-Waveform arrays (as well as in TDMS files), more than 20 attributes, characterizing my measurement conditions.

What I am doing actually, is construct a big (comma delimited) string, (having converted before all my parameters to individual strings) and then use this large string as just one (string) attribute of my waveform.

I haven't found any information, yet, about the optimum way of using many waveform quantities. Does anybody know whether a performance degradation (in speed?, size?) is noticed is such a case ?

Kind regards,

Dimitri

[JÞB · Accepted solution]

I'd avoid the "Super String" Attibute and make them individual attributes:

Attributes of a waveform written to TDMS are automatically added to the file.  Having just one string with all the values is going to make it harder for the end customer to read the file.  Size and Speed should not be issues since the metadata is only written once.  Go for the user friendlier format and use one value per attribute

[crossrulz · Accepted solution]

I'm in full agreement with Jeff here.  You will see little to no difference in performance of writing the super string versus in individual attributes.  So make it easier for the end user (and likely yourself) when trying to read the file.  And if you are using the DataFinder, you will be able to actually do a search on those attributes.  That will really make QA happy.

[Henrik_Volkers · Accepted solution]

Same here, lot of small attributes. Some attribute names are also used by NI,  nice to auto name plots in graphs. 

But I still have one 'super string' attribute called 'history'  that is a log of all signal processing done to the data. human read able 🙂  and it holds the version numbers of the vis that set certain attributes.

BTW: Is there a listing of NI used wfrm attributes ??