kinder@inteloa.intel.com (David B. Kinder) (11/15/88)
AI-420 (submitted over 3 years ago) brought up the issue of an an 8-bit character type. This should be a hot issue with non-US Ada customers who should reasonably insist on support for their (non-English) characters in STRING variables and in TEXT_IO. I've seen packages that define a new extended character type, however this character type can't be used with TEXT_IO, so a whole new (and non-standard) I/O package must be used. Are there Ada users concerned with this problem? Have there been any further thoughts by the LMC/ARG on supporting 8-bit characters? Can we get some definitive ruling on AI-420? What do you Ada developers think about direct language support for 8-bit characters? -- David Kinder, BiiN
sommar@enea.se (Erland Sommarskog) (11/21/88)
David Kinder (kinder@biin.com) writes: >AI-420 (submitted over 3 years ago) brought up the issue of an an 8-bit >character type. This should be a hot issue with non-US Ada customers who >should reasonably insist on support for their (non-English) characters in >STRING variables and in TEXT_IO. I've seen packages that define a new >extended character type, however this character type can't be used with >TEXT_IO, so a whole new (and non-standard) I/O package must be used. I recall I brought up this problem in this newsgroup a little more than a year ago. The conclusion if you wanted to use eight-bit characters was to define a new type with a new Text_IO or to simply cheat. (You are not really required to write an entirely new Text_IO you can map to the old, but you still have problems with being non-standard.) The big problem with Ada is that it defines only characters 32-126 as printable. It also says that a user that in/outputs non-prinable charcaters cannot rely on that Text_IO works as described. I generally very hesitant to proposals to change the Ada standard. One point with Ada was that it should be complete. But a minor change like extending Ada to support the ISO/Latin standards (8859/1-9) has my full support. I even see it as necessary. At minimum 8859/1 should be supported. I don't know about AI-420. Could you, or anyone else, give more info? -- Erland Sommarskog ENEA Data, Stockholm sommar@enea.se "Frequently, unexpected errors are entirely unpredictable" - Digital Equipment