[Info-vax] VSI plans to offer an IA64 to X86 binary translator similar to AEST?

[John Santos]

In article <q6dtrk$u7t$1 at dont-email.me>, 
seaohveh at hoffmanlabs.invalid says...
> 
> On 2019-03-14 14:07:33 +0000, Rod Regier said:
> 
> > Has VSI indicated any plans to offer on OpenVMS IA64 to X86 translater 
> > application...
> 
> Set some expectations with local management...
> 
> Translation is the desperation-pending-app-rewrite part of the migration path.
> 
> That might be a rewrite because of lost source code, or because of a 
> tool that's no longer available from the vendor (and where you have 
> permission to translate).
> 
> The translation path has and likely always will produce apps that are 
> difficult to support and to maintain.
> 
> When translation works, it works.
> 
> When the translated apps crash, figuring out what happened and why?app 
> bug, compiler bug, translation bug, OpenVMS bug?can be difficult, at 
> best.
> 
> Debugging a native app?with the existing and supported architectures, 
> with x86-64 to be determined?is vastly easier than debugging a 
> translated app, too.
> 
> If the application source code is available, then plan to tweak the 
> source code as (if) necessary and recompile and relink, and don't even 
> think about translating.

Not counting issues with build command files (mostly in 
creating shareable images) which required different LINK 
qualifiers and options files, I think about half the 
problems we encountered when porting VAX to Alpha and 
Alpha to I64 were issues with the new compilers 
(ultimately fixed by DEC/Compaq/HP) and about half were 
existing but previously unrecognized bugs 
in our source code.  The latter sort of problems are 
probably uncorrectable by a binary translator.