Re: Bliss History

On 7/12/2024 8:30 AM, Robert A. Brooks wrote:

On 7/12/2024 9:19 AM, Hunter Goatley wrote:
 

I had heard eons ago that DEC had a bunch of BLISS compilers: for
RSTS, for ULTRIX, for Windows. I don't know if they really did, but
they were certainly never made available outside of DEC,
unfortunately.

 There were certainly compilers that generated bliss objects for the PDP-11, but they
were not hosted on a PDP-11; even the mighty, beloved task builder could not make it
fit.
 

An interesting coincidence, I was scanning through my "Downloads" folder, looking for something and came across a file named "bliss.pdf" from Feb 2024.  I don't know where I got it, but it was an article "The BLISS programming language: a history by Ronald F. Brender" ©2002 John Wiley & Sons Ltd.
In it was this paragraph:
By the end of 1979, the new generation of self-hosted and highly-language compatible BLISS compilers was firmly established at DEC. We had BLISS-36 targetting and hosted on the PDP-10, BLISS-32 targetting the VAX-11 and hosted on both the PDP-10 and VAX-11, and BLISS-16 targetting the PDP-11 and hosted on both the PDP-10 and VAX-11.
In 1981 an attempt was made to shoe-horn a variant of BLISS-16 onto the PDP-11 using a combination of language subsetting, somewhat simplified code generation, and heavy overlaying. A constraint was that we were not willing to compromise much on code quality compared to that available from the BLISS-16 cross-compilers. While the resulting compiler generally worked, it was quite limited in the size of module it could compile, because the 65 Kbyte PDP-11 address space was just too small. The compromises were deemed too severe and the experiment was abandoned.
The whole article (it appears to be a clipping of chapter from a larger work) is quite an interesting discussion of BLISS.  If I knew where I had downloaded it from, I'd include the link.  Google did not readily find anything.
Oops.  Never mind.  A little better diffing found this:
https://www.cs.tufts.edu/~nr/cs257/archive/ronald-brender/bliss.pdf
Another quote from near the beginning, guaranteed to make Simon cringe ( :) )
BLISS is a typeless language; that is, all data is manipulated in terms of the underlying machine word sized units. The word size is usually the same as the integer or general purpose register size, which may be a multiple of the addressable unit size.
Enjoy...
--
John H. Reinhardt