Vidéotron wrote: > I have a CS-200 here and I have written some code with this driver. > I don't think it is impossible for Argyll to support the CS-200 but Graeme > insists on have an instrument for testing when writing the code. The problem is threefold: It would be slow to develop a driver remotely, without any access to the instrument. Take the JETI specbos 1211 for instance: - it has a high level, documented communication protocol, yet I probably had to compile code and test and run it many hundreds of times to debug my code and figure out exactly what works and doesn't work with the instrument. Doing that through someone else at one day turnaround (ie. given the time zone differences and time available for running a test and responding), and we're talking months of development and very inefficient use of two peoples time. If it's an undocumented protocol, multiply that by 10 or more, and ramp up the skill needed by the person with access to the instrument. But once a driver is running, there is the problem of keeping it running. I sometimes make wide ranging changes to the driver API, to accommodate some new feature in a way that is orthogonal to the instrument type. Any time I do that, I risk breaking a working driver, so without having an instrument to test with, I simply can't be sure such a driver still works or even runs with each release. Then there is the problem of investigating various problem reports - someone comes back with a "instrument XXX doesn't work on my YYY that uses a ZZZ USB controller". This is hard enough with the instrument, and impossible without it. Graeme Gill.