the mdwn files need to contain all information necessary to convert to latex for inclusion in the Power ISA spec. missing at the moment is the "Fields" specifications, currently listed in fields.text. fields.text ultimately should be autogenerated from the information in the mdwn files but a step-by-step migration path is needed. the first task is therefore to modify the mdwn parser so that it can support reading Field-format the second task is to write a script (not a shell script unless it calls insndb) that inserts Field-format into the mdwn file. the third subtask is to autogenerate the first section of fields.text likely by first splitting it into fields.text and forms.text
(In reply to Luke Kenneth Casson Leighton from comment #0) > the third subtask is to autogenerate the first section of fields.text > likely by first splitting it into fields.text and forms.text one major caveat is that the v3.1B pdf has forms without corresponding fields and/or instructions without corresponding entries in forms/fields, so those need to manually be added if we're hoping to autogenerate that whole section of the pdf. additionally, we'll need hidden fields for at least some of the unnamed bits since iirc they can be more than just /// tokens.
(In reply to Jacob Lifshay from comment #1) > one major caveat is that the v3.1B pdf has forms without corresponding > fields and/or instructions without corresponding entries in forms/fields, not our problem. and i cannot say more about what v3.1D or v3.2 might look like. > so > those need to manually be added if we're hoping to autogenerate that whole > section of the pdf. no - just enough so that the ISA WG can get on with adding them in manually. if they (i mean IBM) choose to use our automated tools GREAT, they can pay their own way doing that.
(In reply to Luke Kenneth Casson Leighton from comment #2) > (In reply to Jacob Lifshay from comment #1) > > so > > those need to manually be added if we're hoping to autogenerate that whole > > section of the pdf. > > no - just enough so that the ISA WG can get on with adding them in > manually. if you're also referring to the hidden fields, we'll need them too if our parser cares about the /// vs. something else distinction.