🚀 go-pugleaf

On Sat, 02 Jun 2018 16:07:09 -0400, Andreas Kohlbach
<a...@spamfence.net> wrote:

>On Sat, 2 Jun 2018 12:23:56 +0100, Gareth's Downstairs Computer wrote:
>>
>> On 02/06/2018 03:44, John Levine wrote:
>>> In article <cb11f87e-44a9-4f0b...@googlegroups.com>,
>>> Quadibloc  <jsa...@ecn.ab.ca> wrote:
>>>
>>> Dartmouth's compiler was quite impressive, typically compiling an
>>> entire program in one or two clock ticks.  It was so fast that nobody
>>> bothered to save object code.
>>
>> Interesting, does that also indicate interactive development where
>> the compilation speed would be largely unapparent?
>
>I guess I am missing a link here. Why did none of the BASICs coming with
>the 8bit micros need a compiler and rather interpret the code instead?

I think the real issue is the nature of a compiler and cost and
availability of hardware.

A compiler takes a source program, generates an object program, and
then to actually do anything you need to load and execute the object
program.

To make this work you need to have enough storage of some kind (paper
tape, mag tape, disk, drum, RAM, doesn't matter, you need to have
enough) to store all the pieces.  Early micros did not have much RAM
and typically if they had a storage device other than the operator's
paper notebook it was paper tape or an audio cassette recorder, so
turnaround time for compiled code would be slow.

An interpreter only needs storage for the source code, a scratchpad,
and the output, so it can run reasonably well on a machine with
limited RAM and a slow storage device, hence the popularity on early
hobbyist micros (and yes, to the nits who think that hobbyists are
only allowed to own toys, I mean the ALTAIR and its clones).