| Re: some users will be harmed by the new licensing policy in IDL 8.6 [message #94074 is a reply to message #94066] |
Fri, 13 January 2017 07:37   |
Patrick Broos
Messages: 27
Registered: December 1996
|
Junior Member |
|
|
Chris,
I appreciate your helpful suggestions, and your obvious desire to support your customers.
With your indulgence, I'd like to describe one of my own astronomy data processing tasks. The telescope I work with (the Chandra X-ray Observatory) observes a target in a series of separate observing "segments", separated in time. Various data processing/analysis computations have to be performed on each of those segments. That segment processing is most naturally and efficiently performed using an IDL session for each segment. On a 12-core machine, > 12 such processes can productively run concurrently (depending on the balance of CPU vs I/O activity). The IDL Bridge is not suitable for this processing, for several reasons.
When that single-segment processing finishes, "merging" computations have to run (to combine results from all the segments). This is most efficiently done by another IDL session, which is launched at the outset, then polls to detect the completion of the segment processing, then performs the merge processing.
While all that number crunching is going on, I may have several interactive IDL sessions in the middle of visualization work on several projects. (Astronomers are always working on several targets and/or proposals at the same time.)
While all that's going on, I may be writing/debugging/testing other IDL programs, which requires another IDL session.
The general theme here is that IDL is integral to almost every part of my daily work activities. I need to keep my 12 cores busy, and that requires many IDL sessions. I have multiple on-going data analysis projects, and it's often very helpful to leave several IDL sessions open on separate desktops for days or weeks.
For 25 years, the monetary cost of this sort of IDL-centric multitasking working style for ONE person has been ONE license. Under the new scheme, this sort of IDL-centric multitasking working style is simply not feasible. I have never even been tempted to jump to Python (as most astronomers are doing), but I feel backed into a corner now.
Sincerely,
Patrick Broos
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
