[enhance@nx2 Data]$
++ 3dTstat: AFNI version=AFNI_2009_12_31_1431 (Feb 25 2010) [64-bit]
++ Authored by: KR Hammett & RW Cox
++ Output dataset ./EPI_All_Runs_mean+orig.BRIK.gz+orig.BRIK
My first thought was that it was just a buggy progress message--that the filename actually being used was correct--but turns out it was telling the truth. I get
EPI_All_Runs_mean+orig.BRIK.gz+orig.HEAD
EPI_All_Runs_mean+orig.BRIK.gz+orig.BRIK
I tried to find out whether that file contained a gzipped dataset or not, but 3dcalc doesn't seem to parse -prefix under the same assumptions that 3dTstat does.
[enhance@nx2 Data]$ 3dinfo EPI_All_Runs_mean+orig.BRIK.gz+orig
++ 3dinfo: AFNI version=AFNI_2009_12_31_1431 (Feb 25 2010) [64-bit]
** FATAL ERROR: Can't open dataset EPI_All_Runs_mean+orig.BRIK.gz+orig
** Program compile date = Feb 25 2010
Why do different binaries have their own idiosyncrasies (i.e. different assumptions) about things like parsing dataset reference strings (whether input or output datasets)? For me at least, it's a regular source of headaches in scripting of processing using AFNI tools. If the code is modularized and the abstractions respected I don't understand the variations in behavior across the suite.
In addition to the particular issue above, if you could help me understand this better I'd appreciate it!
Thanks