Parsing implementation section

This concerns parsing implementation section of a unit, but also parsing a program file (these two things are so much related to each other, it would be trivial to implement one of them having the other). This was already requested a couple of times:

• use also comment in implementation section of code. If you define something (like a class) in implementation section, the comments are not read (or generat ?). This is very important for technical information. –GarfieldFr

• Use comment in interface AND in implementation. Only comment in interface are use. Set comment in implementation can be usefull to set a long comment without destroy the lisibility of interface. ( see PHPDoc ) –GarfieldFr

• Parsing of project file. If you include a project file(*.dpr), there is an error and it is not parsed. A project file can have some code commented. –GarfieldFr

• Reading comments in implementation section was already requested in RFE 1189121, JavaDoc-like Tag handling. Read my comments there – this request is not rejected, but it's quite low priority. – MichalisKamburelis.

• Oh, and it was also requested in RFE 976519, 4 question.... (see 3rd and 4th question). My answers to this were roughly "this will not be implemented" (although I didn't say this absolutely straight there ). – MichalisKamburelis.

My initial answers to this request (see my comments to RFE 1189121 and 976519) we're along the lines "this will not be implemented – too much work, too little gain". But now, considering that 3 different people already requested this, and it's not something unsensible, my answer is: OK, it should be implemented one day.

But still I see more important things to do on pasdoc right now, and it's not a feature that I would use. So let me do a little "call for help" here:

Implementing this thing is mainly a matter of adding to parser (PasDoc_Parser.pas) ability to parse implementation sections of units/programs. The parser need not be a real parser – it can cheat by simply skipping some things without really understanding them (e.g. if you find a token begin then such parser can simply consume tokens up to a matching end, without paying any attention to what it reads). All other things (merging documentation between interface/implementation, some command-line option like --parse-implementation etc.) are trivial.

If anyone would volunteer to implement such thing, please speak up on our MailingList.

I started to work on this feature, if people have input please see the thread Parsing of the implementation section on our MailingList – Mark de Wever