[Info-vax] VAX Macro to C conversion

[Andrew Shaw]

On Tuesday, June 25, 2019 at 5:03:22 AM UTC+10, Stephen Hoffman wrote:

Hi Stephen,

> 
> But the output of most of the automated-conversion converted code 
> around has itself been somewhere less than maintainable.
I appreciate that this is a very real issue that we are going to have to address. Frankly I don't really see the  point doing any conversion whether it be assisted by a tool or completely manual if the end result is difficult to maintain. In that scenario I'd be inclined to leave it where it is.

> 
> And some Macro32 source code can be more complex than a translation 
> tool might permit. OpenVMS has some of that, as is being discussed 
> else-thread.
> 
This is also true and for this I am going to have to do some deeper analysis of the code (as has been pointed out by others in here) in order to understand the subtleties of what is going on - hopefully get an understanding of why it was written in MACRO in the first place. Our system is a curious mix of languages - MACRO-32, C, FORTRAN and RATFOR forms the bulk of the system code. Recent ancilliary / supporting scripts have started appearing in PYTHON as well. As have many existing VMS systems today we have gone through the usual lifecycle of VAX --> ALPHA --> IA64. We do have some extremely time-sensitive requirements in certain aspects of our code (which all runs in memory) and my gut feel is that the MACRO code exists in an effort to streamline those time sensitive bits, but I need to deep dive into that more.

> One common path for these is to have developers convert each module as 
> any substantial modifications are required.
> 
This is the approach I am trying to get some momentum behind. Any time we need to lift the lid on any MACRO code we turn it into C. I guess here I am exploring what tools may exist to help us get that done, rather than just blindly let the tool do it for us. I'm not sure a wholesale migration of all MACRO code to C (or FORTRAN or something else) is necessarily a good approach. I'm a firm believer in the "If it ain't broke don't fix it" approach. And the code does work today. Its worked for close to 30 years.