One of the most important things is the long lifetime of data formats and general programs.
VMS software relies strongly on system-calls which are incompatible with other operating systems.
TRIUMF graphics support for terminal types is limited.
Often not obvious how to export data from VAX program for use on PC. We almost always use a PC for remote access, but facility programs assume a particular local terminal type (e.g. X windows).
Data format must be portable! Please stay away from proprietary software like MS.
To access data with my own programs, I found the MUD library (TRIUMF) convenient. Instead of creating a unified data format, a universal library that can access any data format by the same function calls may be a solution.
The PSI-GPS format uses picoseconds for the time unit, which causes integer-overflow to describe the timing parameters.
Sometimes I am not able to easily transfer data from MINUIT to an Origin data system in a usable form.
UNIX 
VMS: keyboard problems.
TRIUMF program cfit is practically impossible to use unless
one gets his/her Ph.D. at TRIUMF and has years of practice.
Very user unfriendly.
Also requires VAX.
Should have PC format for all analysis
TRIUMF/PSI/ISIS.
Code is not transportable due to different compilers.
See figure Java GUI from roundtable discussion:
http://musr.physics.ubc.ca/fmts/web99_share/web99_share.html
Important that we learn to think in terms of ``passing objects'' rather than ``reading files'' (and agree on specifications of objects!!)
Fitting programs capable of handling both TRIUMF and PSI data formats which are user friendly and don't require converting the initial data files.
No PC version of programs.
Graphics don't work on many different terminals.
Conversion between different formats is probably enough.
Each facility format should be publicly documented and changes prominently announced. Therefore good software should simply autodetect the format, and read the binary files directly, as already done e.g. by Wimda. A common/universal data format is probably not necessary and unlikely to be feasible.
Physics first! Analysis has to follow! It's a must!
It seems quite hard to have a universal data format because counter setups are different. What is highly desirable is to have universal analysis program which allows to treat asymmetries as FBLR -- and standard counter groupings for users who are not from the facilities.
It should be transparent for users how to transfers raw data to analysis programs OR to own storage (i.e. CD, disk, or other.)
For a universal format to be acceptable, perhaps you should consider the following:
The number of data formats is not that large. It will be much more practical for data analysis programs to include the ability to read these few formats rather than expecting all data acquisition systems to converge on one format.
We have to use a secure shell to contact our home systems. Need secure shells for data transfer.
A universal format is certainly necessary, but you would probably want to be able to adjust some machine specific parameters like histograms' background.
A universal fitting program for the time spectra is a very needed tool. It should, however, keep an open character, making it easy to put in a new function in the source code, to be shared by the community.
Simultaneous fitting of several correlated asymmetry spectra would be very useful.
Need to use more than two programs because functions are not always available.
PSI and ISIS fortran binary files are horrid.
For data transfer and results transfer between programs (Mathematica under Windows, Physica under Linux), I have to reboot. I would like:
Universal software such as Francis Pratt's WIMDA is very convenient (even under Softwindows on a Mac), but only so far as it is completely kept up to date.
Convenient output formats for further custom meta-analysis are also important.
Some online software (in particular, all the software at KEK) badly needs updating.
Some keystrokes on our Windows system do not function when logging on to TRIUMF VMS system.
Problems with sending zipped data from one VAX system to LINUX, gets corrupted.
I would be happy if someone found and fixed!! all the bugs in MUD format and msrfit for Linux!!!!! DB seems OK.
Emulation of Tektronics graphics on PCs does not always run well.