| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
P1788 I submit a motion that 1788 Input/Output be as described in the attached draft text of Clause 12. This draft has undergone much comment and revision from (especially) Vincent Lefevre, Michel Hack, Ned Nedialkov and Jürgen WvG. It still needs improvements, especially (I think) in how much "should" and/or "shall" there should be about the conversion specifier cs for Output. But I think it is fairly close to final text. However I emphasise this is a motion on content, not on actual text. So in summary, you are voting that 1. Specifying I/O in 1788 is a good thing. 2. I/O is closely related to interval literals as in Motion 46, and therefore is an important part of the raison d'être of interval literals. 3. An implementation shall have a "conversion specifier", cs, system to say how intervals shall be converted to strings; but the standard leaves it language- or implementation-defined how this is done. As an example, I suggested that in C, a cs like %[%.4f %.6e] might convert [1,2] to the string [1.0000, 2.00000e00] 4. There shall be a documented way to output an interval xx of any type *exactly* as a string s, meaning one can read s back and recover the original xx. 5. 1788 doesn't get involved in stream I/O. So the operations of this clause are like sprintf() rather than fprintf(). I look forward to further constructive comment to help improve the text. Best wishes John Pryce
Attachment:
20130728IOclause.pdf
Description: Adobe PDF document