I fully agree. Except that each charset is 6 lines of code. It's trivial to add them.
________________________________ From: Sloane, Brandon <[email protected]> Sent: Monday, May 20, 2019 6:09:47 PM To: [email protected] Subject: Re: Adding character encoding? I think that is my point. We have a legacy format that defines non-standard charsets that are used no where else, to the point where we are naming the encoding based on the context in which it is used in a particular format. If we want to think of DFDL as a general format description language, then DFDL itself should not have such specific features designed to support specific formats. If I were not so close to daffodil development and I were trying to use it to create link16 schema, I would have come to the conclusion that it was not possible to support the non-standard string encodings. I might have asked the developers to add support for specifying additional encodings; but it would not even occur to me that the developers would be interested in adding a character encoding just to support exactly 1 field in 1 format which they are not able to see. ________________________________ From: Beckerle, Mike <[email protected]> Sent: Monday, May 20, 2019 5:43:34 PM To: [email protected] Subject: Re: Adding character encoding? I think the number and variety of these things is fairly small. They're all about exactly 1 legacy data format, mil-std-6016, so that's not an argument for a generalized facility. I doubt there are many more of them even in that specification. So I suggest we simply add these encodings as we encounter them. ________________________________ From: Sloane, Brandon <[email protected]> Sent: Monday, May 20, 2019 3:31:06 PM To: [email protected] Subject: Re: Adding character encoding? Does it make sense to add support for defining these in schema? The charsets I am adding are pretty speicific; and, questions of public disclosure aside, do not really seem to fit in the global codebase. We can get away with doing this since we happen to be Daffodil Devs, so there is a low barrier to entry for us inserting new encodings. However, if I were just a Daffodil user I would be at a loss for dealing with this usecase. ________________________________ From: Beckerle, Mike <[email protected]> Sent: Monday, May 20, 2019 12:57:41 PM To: [email protected] Subject: Re: Adding character encoding? The encodings are in daffodil-io, and for these small charsets it's very very easy to support them. There is already a 6 bit charset for DFI 264 DUI 001, and also some 5-bit encodings. All this was motivated by Link16. Take a look at file X_DFDL_6_BIT_DFI_264_DUI_001.scala ________________________________ From: Sloane, Brandon <[email protected]> Sent: Monday, May 20, 2019 11:43:40 AM To: [email protected] Subject: Adding character encoding? In working on schema for mil-std-6016, I have discovered a 6 bit character encoding that is distinct from the X-DFDL-US-ASCII-6-BIT-PACKED encoding we already support. What would the process for supporting this be? Brandon T. Sloane Associate, Services [email protected] | tresys.com
