FYI TDMS open and read on empty file gives no error's out

Question

I found that if you 'open' an empty (0 byte) file using TDMS open, you will not get any error messages when you open it, nor when you call functions such as "Get properties" or "List content", also no error when you close the TDMS file.
If the file is non-empty, but not a valid TDMS file, TDMS open returns an error.
I would have expected that I would get an error or a warning at least when trying to read a non-existant property or group, if not by the open function itself?!
The 2013 snippet below creates a 1-byte or empty file in the local user temporary folder, closes the file, then attempts to open and read it as a TDMS file.  The temporary file is deleted afterwards.
If the file is empty, NO errors or warnings occur at any point.  If the file is not empty, it fails on the TDMS open. (Note the sequence structure is because the path out of the tdms close is invalid if the tdms open failed.
This could conceivably cause an issue if your program does not do additional fault checking to catch existing but empty files.  (Which of course shouldn't ever happen, but in some situations it could happen.)
QFang
CLD LabVIEW 7.1 to 2013

Answer 1

Hooovahh
I asked to list the channels in a named group.  If that named group did not exist, I would think that should at least be a warning,  but I could see how it should not be an error, it was just unexpected to me that reading a non existant group (in an empty file) did not indicate any issues at all.   To me, 'error out' should give a warning or an error if something unexpected happened, but I may just have stricter opinion on something that by nature is very philosophical and open for debate. (And there are always tradeoffs, overt checks and bounds in a function obviously leads to overhead, and TDMS is optimized for performance, so that may be one reason it is the way it is.)
(This situation came up for me because when a file is first created, we datetime tag it and write a bunch of other meta-data to the file, then later (and over time) we add the spectrum goup + data.  [Edit] actually, in the old code, if the file existed, it was assumed to have been setup correctly. so if the group exists, the old code assumed all the meta data has already been written [end edit].. One of the many fixes possible, I'll rely on the 'found' boolean output of the parameter read instead of relying on the error/warning. If the datetime tag is found, the code will asume the file exists and has been filled with the meta data I expect.)
Thoult ::  I wasn't sure, but in either case, my intent was to provide extra information, not 'knock you' for participating in the discussion, I hope thats not how I came accross.  English is a second language to me, so I sometimes come across 'wrong' from what I mean.
QFang
CLD LabVIEW 7.1 to 2013