Looking for Johnson's pcc - DG Nova and Eclipse - Unix

This is a discussion on Looking for Johnson's pcc - DG Nova and Eclipse - Unix ; Hi, In their paper, "Portability of C Programs and the UNIX System", S.C. Johnson and Dennis Ritchie write, \\ Even before the idea of moving UNIX occurred to us, it was clear that C was successful enough to warrant production ...

+ Reply to Thread
Page 1 of 2 1 2 LastLast
Results 1 to 20 of 36

Thread: Looking for Johnson's pcc - DG Nova and Eclipse

  1. Looking for Johnson's pcc - DG Nova and Eclipse

    Hi,

    In their paper, "Portability of C Programs and the UNIX System", S.C.
    Johnson and Dennis Ritchie write,

    \\
    Even before the idea of moving UNIX occurred to us, it was clear that C
    was successful enough to warrant production of compilers for an
    increasing variety of machines. Therefore, one of the authors (SCJ)
    undertook to produce a new compiler intended from the start to be
    easily modified. This new compiler is now in use on the IBM System/370
    under both OS and TSS, the Honeywell 6000, the
    Interdata 8/32, the SEL86 the Data General Nova and Eclipse, the DEC
    VAX11/780, and a Bell System processor. Versions are in progress for
    the Intel 8086 microprocessor and other machines.
    //

    Do any of these ports of pcc or pcc2 still exist? I am interested in
    the Nova/Eclipse versions in particular.

    --Toby


  2. Re: Looking for Johnson's pcc - DG Nova and Eclipse


    "toby" wrote in message news:1145627720.914581.63620@v46g2000cwv.googlegro ups.com...

    Toby asked about versions of Johnson's pcc, in particular
    >
    > Do any of these ports of pcc or pcc2 still exist? I am interested in
    > the Nova/Eclipse versions in particular.
    >
    > --Toby


    I'm not aware of extant versions of Nova/Eclipse versions of this.
    pcc versions for the PDP11 were in Seventh Ed. and for the VAX
    in 32V as well as the earlier BSD distributions (these are definitely
    available at tuhs.org). The other instances would be harder to find.
    Early Sun systems used it for the Moto 68K, for example.

    Dennis



  3. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    On Sat, 22 Apr 2006 03:50:09 -0000, "Dennis Ritchie"
    wrote:

    >
    >"toby" wrote in message news:1145627720.914581.63620@v46g2000cwv.googlegro ups.com...
    >
    >Toby asked about versions of Johnson's pcc, in particular
    >>
    >> Do any of these ports of pcc or pcc2 still exist? I am interested in
    >> the Nova/Eclipse versions in particular.
    >>
    >> --Toby

    >
    >I'm not aware of extant versions of Nova/Eclipse versions of this.
    >pcc versions for the PDP11 were in Seventh Ed. and for the VAX
    >in 32V as well as the earlier BSD distributions (these are definitely
    >available at tuhs.org). The other instances would be harder to find.
    >Early Sun systems used it for the Moto 68K, for example.
    >

    I would be interested in the Nova version, if a copy ever surfaces.
    16-bit word machine, 4 regs, no stack. . . The Eclipse was a somewhat
    different critter, and I never saw one.

    Might be fun to try to port 6th or 7th Edition to a Nova. :-) It
    would give me a reason to get the Nova clones I have working again.

    --
    ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    BCET Basic Compiler Page: http://www.arargh.com/basic/index.html

    To reply by email, remove the garbage from the reply address.

  4. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    wrote in message
    news:elbj42t9fica9og6shnjo5deihfnb6ml5h@4ax.com...
    > On Sat, 22 Apr 2006 03:50:09 -0000, "Dennis Ritchie"
    > I would be interested in the Nova version, if a copy ever surfaces.
    > 16-bit word machine, 4 regs, no stack. . .


    Later versions had both a stack and a frame pointer (thought I expect you knew
    that).

    Didn't help me any, though, since my code had to run on the entire product line
    I couldn't use them.

    - Bill



  5. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    > On Sat, 22 Apr 2006 03:50:09 -0000, "Dennis Ritchie"
    > wrote:
    >
    > >
    > >"toby" wrote in message news:1145627720.914581.63620@v46g2000cwv.googlegro ups.com...
    > >
    > >Toby asked about versions of Johnson's pcc, in particular
    > >>
    > >> Do any of these ports of pcc or pcc2 still exist? I am interested in
    > >> the Nova/Eclipse versions in particular.
    > >>
    > >> --Toby

    > >
    > >I'm not aware of extant versions of Nova/Eclipse versions of this.
    > >pcc versions for the PDP11 were in Seventh Ed. and for the VAX
    > >in 32V as well as the earlier BSD distributions (these are definitely
    > >available at tuhs.org). The other instances would be harder to find.
    > >Early Sun systems used it for the Moto 68K, for example.
    > >

    > I would be interested in the Nova version, if a copy ever surfaces.
    > 16-bit word machine, 4 regs, no stack. . . The Eclipse was a somewhat
    > different critter, and I never saw one.
    >
    > Might be fun to try to port 6th or 7th Edition to a Nova. :-)


    Yes, it would... If only it were still around, this compiler would be a
    nice start. iirc Johnson's "pcc2" was an improved version that made
    retargetability easier (I can't exactly recall where I read this).
    Perhaps, starting from pcc2, it would be possible to simply redo
    creating the Nova/Eclipse target. (Although the original work probably
    required many changes to the compiler itself beyond the machine
    description, since target parameters would not likely have included
    support for 2 pointer types, etc.)

    Was the original pcc/DG port part of a UNIX port? Or was it intended to
    run under RDOS or similar? Chris Torek's past posts often refer to C on
    Nova/Eclipse, but I suspect he wasn't talking about pcc itself.

    --Toby

    > It would give me a reason to get the Nova clones I have working again.
    >
    > --
    > ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    > BCET Basic Compiler Page: http://www.arargh.com/basic/index.html
    >
    > To reply by email, remove the garbage from the reply address.



  6. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    On Sat, 22 Apr 2006 02:17:07 -0400, "William J. Leary Jr."
    wrote:

    > wrote in message
    >news:elbj42t9fica9og6shnjo5deihfnb6ml5h@4ax.com...


    >> I would be interested in the Nova version, if a copy ever surfaces.
    >> 16-bit word machine, 4 regs, no stack. . .

    >
    >Later versions had both a stack and a frame pointer (thought I expect you knew
    >that).


    I thought only Eclipses' had those. But, I really don't know as I
    never used any Nova, except for my DCC-116s. I think that they are
    Nova 2 clones, but a bit faster.

    >Didn't help me any, though, since my code had to run on the entire product line
    >I couldn't use them.

    --
    ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    BCET Basic Compiler Page: http://www.arargh.com/basic/index.html

    To reply by email, remove the garbage from the reply address.

  7. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    wrote in message
    news:mi2l429q3kjd4v95mjai2gvborplm4588m@4ax.com...
    > On Sat, 22 Apr 2006 02:17:07 -0400, "William J. Leary Jr."
    > wrote:
    > > wrote in message
    > >news:elbj42t9fica9og6shnjo5deihfnb6ml5h@4ax.com...
    > >> I would be interested in the Nova version, if a copy ever surfaces.
    > >> 16-bit word machine, 4 regs, no stack. . .

    > >
    > >Later versions had both a stack and a frame pointer (thought I expect you

    knew
    > >that).

    >
    > I thought only Eclipses' had those.


    Nova 3 added a stack and fram pointer. It's arguable that the Nova 3 wanted to
    be an Eclipse. It could at least be viewed as a stepping stone in that
    direction.

    > But, I really don't know as I
    > never used any Nova, except for my DCC-116s. I think that they are
    > Nova 2 clones, but a bit faster.


    The DCC-116 was a Nova 2 clone, as I recall. The 216, 316, etc. were faster
    versions of the same archetecture.

    - Bill



  8. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    On 22 Apr 2006 08:17:07 -0700, "toby"
    wrote:

    >ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    >> On Sat, 22 Apr 2006 03:50:09 -0000, "Dennis Ritchie"
    >> wrote:
    >> >"toby" wrote in message news:1145627720.914581.63620@v46g2000cwv.googlegro ups.com...
    >> >Toby asked about versions of Johnson's pcc, in particular
    >> >> Do any of these ports of pcc or pcc2 still exist? I am interested in
    >> >> the Nova/Eclipse versions in particular.
    >> >I'm not aware of extant versions of Nova/Eclipse versions of this.
    >> >pcc versions for the PDP11 were in Seventh Ed. and for the VAX
    >> >in 32V as well as the earlier BSD distributions (these are definitely
    >> >available at tuhs.org). The other instances would be harder to find.
    >> >Early Sun systems used it for the Moto 68K, for example.
    >> >

    >> I would be interested in the Nova version, if a copy ever surfaces.
    >> 16-bit word machine, 4 regs, no stack. . . The Eclipse was a somewhat
    >> different critter, and I never saw one.
    >>
    >> Might be fun to try to port 6th or 7th Edition to a Nova. :-)

    >
    >Yes, it would... If only it were still around, this compiler would be a
    >nice start. iirc Johnson's "pcc2" was an improved version that made
    >retargetability easier (I can't exactly recall where I read this).
    >Perhaps, starting from pcc2, it would be possible to simply redo
    >creating the Nova/Eclipse target. (Although the original work probably
    >required many changes to the compiler itself beyond the machine
    >description, since target parameters would not likely have included
    >support for 2 pointer types, etc.)


    Well, I have the pcc source from 7th Ed., and if I got interested
    enough, I could always tackle porting it to the Nova. I wonder if I
    could use the auto-increment/auto-decrement low core locations as a
    fake stack pointer? I would also have to use R2 as a frame pointer.

    >Was the original pcc/DG port part of a UNIX port?

    I have no idea.

    >Or was it intended to
    >run under RDOS or similar? Chris Torek's past posts often refer to C on
    >Nova/Eclipse, but I suspect he wasn't talking about pcc itself.

    --
    ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    BCET Basic Compiler Page: http://www.arargh.com/basic/index.html

    To reply by email, remove the garbage from the reply address.

  9. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    On Sat, 22 Apr 2006 16:29:05 -0400, "William J. Leary Jr."
    wrote:

    > wrote in message
    >news:mi2l429q3kjd4v95mjai2gvborplm4588m@4ax.com...


    >> I thought only Eclipses' had those.

    >
    >Nova 3 added a stack and fram pointer. It's arguable that the Nova 3 wanted to
    >be an Eclipse. It could at least be viewed as a stepping stone in that
    >direction.


    Ok. Never used a Nova 3.

    >> But, I really don't know as I
    >> never used any Nova, except for my DCC-116s. I think that they are
    >> Nova 2 clones, but a bit faster.

    >
    >The DCC-116 was a Nova 2 clone, as I recall. The 216, 316, etc. were faster
    >versions of the same archetecture.

    Nova 1200 clone, according to the URL below.


    I have 4 DCC D-116 Es, and I once saw a DCC-616. A FPOE had one in
    for an eval. 1975 maybe? I think that DCC folded (or rather got
    folded) about then. I don't know if the 616 ever made it into
    production. I doubt it.

    Something of a DCC mini history:
    http://www.simulogics.com/nostalgia/DCC/dcc.htm
    --
    ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    BCET Basic Compiler Page: http://www.arargh.com/basic/index.html

    To reply by email, remove the garbage from the reply address.

  10. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    wrote in message
    news:rh4l42llni2dnv6dsq45d2ngd0gma8sp26@4ax.com...
    > On Sat, 22 Apr 2006 16:29:05 -0400, "William J. Leary Jr."
    > wrote:
    > >The DCC-116 was a Nova 2 clone, as I recall. The 216, 316, etc. were faster
    > >versions of the same archetecture.

    > Nova 1200 clone, according to the URL below.


    Right... typo.

    The Nova 1200 was so-named because that was it's cycle time. If you happen to
    know the DCC-116's cycle time, that'll tell if it was faster or not.

    > I have 4 DCC D-116 Es, and I once saw a DCC-616. A FPOE had one in
    > for an eval. 1975 maybe? I think that DCC folded (or rather got
    > folded) about then. I don't know if the 616 ever made it into
    > production. I doubt it.


    I used a couple of DCC's, running RDOS. I *think* the 116 version, but I can't
    say from this point in time.

    As I recalled, we got them through some deal with DG after DG sued DCC out of
    business and absorbed their stock. If I'm remembering the deal, we were
    allowed to use them for internal operations (test beds and so forth), but
    couldn't include them in our OEM product (which was built around Nova 1200's,
    800's, and 3's).

    > Something of a DCC mini history:
    > http://www.simulogics.com/nostalgia/DCC/dcc.htm


    Ah, yes. There's part of that story there.

    - Bill



  11. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    On Sat, 22 Apr 2006 17:39:36 -0400, "William J. Leary Jr."
    wrote:

    > wrote in message
    >news:rh4l42llni2dnv6dsq45d2ngd0gma8sp26@4ax.com...
    >> On Sat, 22 Apr 2006 16:29:05 -0400, "William J. Leary Jr."
    >> wrote:
    >> >The DCC-116 was a Nova 2 clone, as I recall. The 216, 316, etc. were faster
    >> >versions of the same archetecture.

    >> Nova 1200 clone, according to the URL below.

    >
    >Right... typo.
    >
    >The Nova 1200 was so-named because that was it's cycle time. If you happen to
    >know the DCC-116's cycle time, that'll tell if it was faster or not.

    980ns is the number I remember. IIRC, the ones I have are D-116EH's
    (extended chassis & faster clock). The DCC quick reference has no
    timings, and the Nixdorf version of the card has two different times,
    800 & 1200. I guess I would have to check the prints & crystals to
    find out just how fast the ones I have really are.

    >> I have 4 DCC D-116 Es, and I once saw a DCC-616. A FPOE had one in
    >> for an eval. 1975 maybe? I think that DCC folded (or rather got
    >> folded) about then. I don't know if the 616 ever made it into
    >> production. I doubt it.

    >
    >I used a couple of DCC's, running RDOS. I *think* the 116 version, but I can't
    >say from this point in time.
    >
    >As I recalled, we got them through some deal with DG after DG sued DCC out of
    >business and absorbed their stock. If I'm remembering the deal, we were
    >allowed to use them for internal operations (test beds and so forth), but
    >couldn't include them in our OEM product (which was built around Nova 1200's,
    >800's, and 3's).

    I don't know where Nixdorf got them, but they were selling DCC's into
    the early 80's as the processor in the 8870 system. Later, they
    started building processors in Germany, which weren't all that
    compatible, especially in the I/O department.


    --
    ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    BCET Basic Compiler Page: http://www.arargh.com/basic/index.html

    To reply by email, remove the garbage from the reply address.

  12. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    > On 22 Apr 2006 08:17:07 -0700, "toby"
    > wrote:
    >
    > >ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    > >> On Sat, 22 Apr 2006 03:50:09 -0000, "Dennis Ritchie"
    > >> wrote:
    > >> >"toby" wrote in message news:1145627720.914581.63620@v46g2000cwv.googlegro ups.com...
    > >> >Toby asked about versions of Johnson's pcc, in particular
    > >> >> Do any of these ports of pcc or pcc2 still exist? I am interested in
    > >> >> the Nova/Eclipse versions in particular.
    > >> >I'm not aware of extant versions of Nova/Eclipse versions of this.
    > >> >pcc versions for the PDP11 were in Seventh Ed. and for the VAX
    > >> >in 32V as well as the earlier BSD distributions (these are definitely
    > >> >available at tuhs.org). The other instances would be harder to find.
    > >> >Early Sun systems used it for the Moto 68K, for example.
    > >> >
    > >> I would be interested in the Nova version, if a copy ever surfaces.
    > >> 16-bit word machine, 4 regs, no stack. . . The Eclipse was a somewhat
    > >> different critter, and I never saw one.
    > >>
    > >> Might be fun to try to port 6th or 7th Edition to a Nova. :-)

    > >
    > >Yes, it would... If only it were still around, this compiler would be a
    > >nice start. iirc Johnson's "pcc2" was an improved version that made
    > >retargetability easier (I can't exactly recall where I read this).
    > >Perhaps, starting from pcc2, it would be possible to simply redo
    > >creating the Nova/Eclipse target. (Although the original work probably
    > >required many changes to the compiler itself beyond the machine
    > >description, since target parameters would not likely have included
    > >support for 2 pointer types, etc.)

    >
    > Well, I have the pcc source from 7th Ed., and if I got interested
    > enough, I could always tackle porting it to the Nova.


    Is that pcc1 or pcc2 (I'm going from vague memory here from some
    reading a year or so ago, that pcc went through an improved rewrite)?

    > I wonder if I
    > could use the auto-increment/auto-decrement low core locations as a
    > fake stack pointer? I would also have to use R2 as a frame pointer.


    I put quite a bit of thought into targeting lcc[1] to Nova, but this
    foundered mainly on these points:
    * an essential presumption of byte addressing means modifying lcc
    itself to deal with the concept of 2 pointer classes;
    * lcc much prefers many more registers (even the PDP-11[2] strains this
    to the limit);
    * supporting long types is not particularly elegant in lcc 16-bit
    machine descriptions.

    Nonetheless, Nova is such a fun architecture that the challenge of a
    producing a good C for it isn't easily ignored. Simh makes a good
    environment for testing under RDOS, and is probably good enough to test
    a UNIX port in, also

    [1] http://www.cs.princeton.edu/software/lcc/
    [2] http://telegraphics.com.au/sw/info/lcc-pdp11.html

    >
    > >Was the original pcc/DG port part of a UNIX port?

    > I have no idea.
    >
    > >Or was it intended to
    > >run under RDOS or similar? Chris Torek's past posts often refer to C on
    > >Nova/Eclipse, but I suspect he wasn't talking about pcc itself.

    > --
    > ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    > BCET Basic Compiler Page: http://www.arargh.com/basic/index.html
    >
    > To reply by email, remove the garbage from the reply address.



  13. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    On 22 Apr 2006 23:07:33 -0700, "toby"
    >ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    >> Well, I have the pcc source from 7th Ed., and if I got interested
    >> enough, I could always tackle porting it to the Nova.

    >
    >Is that pcc1 or pcc2 (I'm going from vague memory here from some
    >reading a year or so ago, that pcc went through an improved rewrite)?

    I have no idea.

    >> I wonder if I
    >> could use the auto-increment/auto-decrement low core locations as a
    >> fake stack pointer? I would also have to use R2 as a frame pointer.

    >
    >I put quite a bit of thought into targeting lcc[1] to Nova, but this
    >foundered mainly on these points:
    >* an essential presumption of byte addressing means modifying lcc
    >itself to deal with the concept of 2 pointer classes;

    That problem would exist for any word addressed system.

    >* lcc much prefers many more registers (even the PDP-11[2] strains this
    >to the limit);

    I would be more inclined to start with something like the 'Small-C'
    compiler, and work up.

    this one:
    Small-C Compiler
    Copyright 1982, 1983, 1985, 1988 J. E. Hendrix


    >* supporting long types is not particularly elegant in lcc 16-bit
    >machine descriptions.

    All I would tackle is char, int, long (8,16,32) -- and maybe skip long

    >
    >Nonetheless, Nova is such a fun architecture that the challenge of a
    >producing a good C for it isn't easily ignored. Simh makes a good
    >environment for testing under RDOS, and is probably good enough to test
    >a UNIX port in, also
    >
    >[1] http://www.cs.princeton.edu/software/lcc/
    >[2] http://telegraphics.com.au/sw/info/lcc-pdp11.html
    >


    --
    ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    BCET Basic Compiler Page: http://www.arargh.com/basic/index.html

    To reply by email, remove the garbage from the reply address.

  14. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    > On 22 Apr 2006 23:07:33 -0700, "toby"
    > >ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    > >> Well, I have the pcc source from 7th Ed., and if I got interested
    > >> enough, I could always tackle porting it to the Nova.

    > >
    > >Is that pcc1 or pcc2 (I'm going from vague memory here from some
    > >reading a year or so ago, that pcc went through an improved rewrite)?

    > I have no idea.
    >
    > >> I wonder if I
    > >> could use the auto-increment/auto-decrement low core locations as a
    > >> fake stack pointer? I would also have to use R2 as a frame pointer.

    > >
    > >I put quite a bit of thought into targeting lcc[1] to Nova, but this
    > >foundered mainly on these points:
    > >* an essential presumption of byte addressing means modifying lcc
    > >itself to deal with the concept of 2 pointer classes;

    > That problem would exist for any word addressed system.


    Indeed. I expect pcc made the same presumption.

    >
    > >* lcc much prefers many more registers (even the PDP-11[2] strains this
    > >to the limit);

    > I would be more inclined to start with something like the 'Small-C'
    > compiler, and work up.
    >
    > this one:
    > Small-C Compiler
    > Copyright 1982, 1983, 1985, 1988 J. E. Hendrix
    >
    >
    > >* supporting long types is not particularly elegant in lcc 16-bit
    > >machine descriptions.

    > All I would tackle is char, int, long (8,16,32) -- and maybe skip long


    I was referring to long = 32. In an lcc machine description this gets
    ugly for PDP-11; another of its presumptions is that you'll have
    registers of your operand width (which for longs is true on any 32 bit
    system of course). Just another indication that while it's possible to
    go there, 16 bitters are outside lcc's sweet spot. I agree other
    compilers are probably easier starting points (I wonder about tcc[3]?).
    lcc's advantage, apart from anything else, is it is correct and
    complete ANSI.

    [3] http://fabrice.bellard.free.fr/tcc/

    > ...
    > --
    > ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    > BCET Basic Compiler Page: http://www.arargh.com/basic/index.html
    >
    > To reply by email, remove the garbage from the reply address.



  15. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    On 23 Apr 2006 10:45:08 -0700, "toby"
    wrote:

    >ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    >> On 22 Apr 2006 23:07:33 -0700, "toby"
    >> >ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    >> >> Well, I have the pcc source from 7th Ed., and if I got interested
    >> >> enough, I could always tackle porting it to the Nova.
    >> >
    >> >Is that pcc1 or pcc2 (I'm going from vague memory here from some
    >> >reading a year or so ago, that pcc went through an improved rewrite)?

    >> I have no idea.
    >>
    >> >> I wonder if I
    >> >> could use the auto-increment/auto-decrement low core locations as a
    >> >> fake stack pointer? I would also have to use R2 as a frame pointer.
    >> >
    >> >I put quite a bit of thought into targeting lcc[1] to Nova, but this
    >> >foundered mainly on these points:
    >> >* an essential presumption of byte addressing means modifying lcc
    >> >itself to deal with the concept of 2 pointer classes;

    >> That problem would exist for any word addressed system.

    >
    >Indeed. I expect pcc made the same presumption.
    >
    >>
    >> >* lcc much prefers many more registers (even the PDP-11[2] strains this
    >> >to the limit);

    >> I would be more inclined to start with something like the 'Small-C'
    >> compiler, and work up.
    >>
    >> this one:
    >> Small-C Compiler
    >> Copyright 1982, 1983, 1985, 1988 J. E. Hendrix
    >>
    >>
    >> >* supporting long types is not particularly elegant in lcc 16-bit
    >> >machine descriptions.

    >> All I would tackle is char, int, long (8,16,32) -- and maybe skip long

    >
    >I was referring to long = 32. In an lcc machine description this gets
    >ugly for PDP-11; another of its presumptions is that you'll have
    >registers of your operand width (which for longs is true on any 32 bit
    >system of course). Just another indication that while it's possible to
    >go there, 16 bitters are outside lcc's sweet spot. I agree other
    >compilers are probably easier starting points (I wonder about tcc[3]?).
    >lcc's advantage, apart from anything else, is it is correct and
    >complete ANSI.
    >
    >[3] http://fabrice.bellard.free.fr/tcc/
    >

    tcc looks a little too 32BITish and a little too X86ish for my taste,
    after a very quick look.

    The reason I would start with a Small-C Compiler is that it is already
    a 16 BITter, and the fact that I have the source and a working
    version.

    Actually I seem to have several versions of 'Small-C' compilers. The
    one I mentioned above is not the one that I was thinking of when I
    wrote the following paragraph. The version I was thinking of is:
    "small-c:PC compiler by Ron Cain"

    It's main problem is that it generates fairly ugly code, at least the
    x86 version does. I could probably hack a version of this in a week
    or two, and get it to output Nova assembler code. Maybe I will, just
    for something to do.

    Hmmm, size may be an issue. The MS/PC DOS exe is 40k bytes - that's
    over half of the Nova's memory. :-)

    Another issue, do I write it for the only Nova OS that I have (IRIS
    from EDS, later Point4) or standalone?

    Also, write for a stock Nova, or the modified clone processors that I
    have. The ones I have support 64k words of core, but you lose the
    LDA/STA memory indirect chaining feature. But they will run as stock.

    --
    ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    BCET Basic Compiler Page: http://www.arargh.com/basic/index.html

    To reply by email, remove the garbage from the reply address.

  16. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    > On 23 Apr 2006 10:45:08 -0700, "toby"
    > wrote:
    > ...I agree other
    > >compilers are probably easier starting points (I wonder about tcc[3]?).
    > >lcc's advantage, apart from anything else, is it is correct and
    > >complete ANSI.
    > >
    > >[3] http://fabrice.bellard.free.fr/tcc/
    > >

    > tcc looks a little too 32BITish and a little too X86ish for my taste,
    > after a very quick look.
    >
    > The reason I would start with a Small-C Compiler is that it is already
    > a 16 BITter, and the fact that I have the source and a working
    > version.
    >
    > Actually I seem to have several versions of 'Small-C' compilers. The
    > one I mentioned above is not the one that I was thinking of when I
    > wrote the following paragraph. The version I was thinking of is:
    > "small-c:PC compiler by Ron Cain"
    >
    > It's main problem is that it generates fairly ugly code, at least the
    > x86 version does. I could probably hack a version of this in a week
    > or two, and get it to output Nova assembler code. Maybe I will, just
    > for something to do.


    You can use my cross-assembler[4] for the next stage :-) It generates
    RDOS relocatable object (RB). Do you have reference manuals for the
    assembler syntax? I can certainly provide them.

    >
    > Hmmm, size may be an issue. The MS/PC DOS exe is 40k bytes - that's
    > over half of the Nova's memory. :-)


    Cross-compilation/cross-assembly is fine for my purposes.

    >
    > Another issue, do I write it for the only Nova OS that I have (IRIS
    > from EDS, later Point4) or standalone?


    Ooh! Please tell me more about that O/S. What are the chances of
    getting a disk image to run under Simh? Do you have documentation? The
    only Nova O/S that I have seen is RDOS (an image ships with Simh).

    >
    > Also, write for a stock Nova, or the modified clone processors that I
    > have. The ones I have support 64k words of core, but you lose the
    > LDA/STA memory indirect chaining feature. But they will run as stock.


    My main interest is Nova3 (I have a friend with two of them).

    [4] http://www.telegraphics.com.au/sw/#dpa

    >
    > --
    > ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    > BCET Basic Compiler Page: http://www.arargh.com/basic/index.html
    >
    > To reply by email, remove the garbage from the reply address.



  17. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    >
    > Actually I seem to have several versions of 'Small-C' compilers. The
    > one I mentioned above is not the one that I was thinking of when I
    > wrote the following paragraph. The version I was thinking of is:
    > "small-c:PC compiler by Ron Cain"
    >
    > It's main problem is that it generates fairly ugly code, at least the
    > x86 version does. I could probably hack a version of this in a week
    > or two, and get it to output Nova assembler code. Maybe I will, just
    > for something to do.


    It couldn't do much else, given the hardware it was originally targeted
    to run on. Also, perhaps it would do better on a less CISC machine than
    the x86.
    ....
    >
    > Another issue, do I write it for the only Nova OS that I have (IRIS
    > from EDS, later Point4) or standalone?


    Is it possible to generate code that will run either way, and have the
    OS-dependent (or standalone-dependent) stuff in the RTL?


  18. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    On 23 Apr 2006 12:17:55 -0700, "toby"
    >
    >You can use my cross-assembler[4] for the next stage :-) It generates
    >RDOS relocatable object (RB). Do you have reference manuals for the
    >assembler syntax? I can certainly provide them.

    I can probably remember it.


    >> Hmmm, size may be an issue. The MS/PC DOS exe is 40k bytes - that's
    >> over half of the Nova's memory. :-)

    >
    >Cross-compilation/cross-assembly is fine for my purposes.

    If I ever get that far.

    >> Another issue, do I write it for the only Nova OS that I have (IRIS
    >> from EDS, later Point4) or standalone?

    >
    >Ooh! Please tell me more about that O/S.

    Besides that it's long dead :-)

    Multi user time sharing system. Apps programmed in interpreted basic.
    System written in assembler.


    >What are the chances of
    >getting a disk image to run under Simh?

    Slim, for several reasons. It is still under copyright, and has never
    been released, AFAIK. Also, in order for it to run, there is a
    license device, called a 'PICO', that has to be fastened to the
    backplane. And they are version specific. The same kind of thing as
    used on the parallel port for software protection for PC's.

    Another issue is that all the copies that I have are setup for either
    Diablo 44 Drives, on the DCC controller (Device ID 33) or Point 4 SMD
    controller (Device ID 27). I doubt that Simh knows about those
    controllers.

    >Do you have documentation?

    Of course.

    >The only Nova O/S that I have seen is RDOS (an image ships with Simh).

    ISTR that DG made RDOS avail for personal use. I don't know that to
    be true for IRIS. That doesn't matter for me, as I have the licenses,
    but it does for allowing anyone else to have a copy. I have
    corresponded with someone who was trying to get IRIS released. I will
    contact him again to see if he has made any progress. It has been a
    few years.

    >> Also, write for a stock Nova, or the modified clone processors that I
    >> have. The ones I have support 64k words of core, but you lose the
    >> LDA/STA memory indirect chaining feature. But they will run as stock.

    >
    >My main interest is Nova3 (I have a friend with two of them).


    Well, since the Nova 3 has a stack, life would be much easier for a C
    compiler. But, I don't have a Nova 3. I knew someone that had one, I
    wonder if he still does. Maybe he would give it to me, if he no
    longer wants it. Have to check.

    --
    ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    BCET Basic Compiler Page: http://www.arargh.com/basic/index.html

    To reply by email, remove the garbage from the reply address.

  19. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    >> ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    >>

    .... snip ...
    >>>
    >>> this one:
    >>> Small-C Compiler
    >>> Copyright 1982, 1983, 1985, 1988 J. E. Hendrix
    >>>

    .... snip ...
    >
    > Actually I seem to have several versions of 'Small-C' compilers.
    > The one I mentioned above is not the one that I was thinking of
    > when I wrote the following paragraph. The version I was thinking
    > of is: "small-c:PC compiler by Ron Cain"


    Ron Cain was the originator, who then retired from the field. I
    always felt that the seizure by Hendrix was not right, although he
    did make improvements.

    --
    "If you want to post a followup via groups.google.com, don't use
    the broken "Reply" link at the bottom of the article. Click on
    "show options" at the top of the article, then click on the
    "Reply" at the bottom of the article headers." - Keith Thompson
    More details at:
    Also see



  20. Re: Looking for Johnson's pcc - DG Nova and Eclipse

    On Sun, 23 Apr 2006 16:13:40 -0400, CBFalconer
    wrote:

    >ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    >>> ArarghMail604NOSPAM@NOT.AT.Arargh.com wrote:
    >>>

    >... snip ...
    >>>>
    >>>> this one:
    >>>> Small-C Compiler
    >>>> Copyright 1982, 1983, 1985, 1988 J. E. Hendrix
    >>>>

    >... snip ...
    >>
    >> Actually I seem to have several versions of 'Small-C' compilers.
    >> The one I mentioned above is not the one that I was thinking of
    >> when I wrote the following paragraph. The version I was thinking
    >> of is: "small-c:PC compiler by Ron Cain"

    >
    >Ron Cain was the originator, who then retired from the field.

    Any idea why?

    >I always felt that the seizure by Hendrix was not right, although he
    >did make improvements.

    The only noticeable improvement that I remember is a peep-hole
    optimizer of sorts, in version 2.2. There may have been others.

    I played around mostly with the 1.1 version: "small-c:PC compiler by
    Ron Cain", and am most familiar with it.
    --
    ArarghMail604 at [drop the 'http://www.' from ->] http://www.arargh.com
    BCET Basic Compiler Page: http://www.arargh.com/basic/index.html

    To reply by email, remove the garbage from the reply address.

+ Reply to Thread
Page 1 of 2 1 2 LastLast