Re: Strange question

Re: Strange question

[Robert Dewar]

<<I'd like to ask you a strange question: how many time
did it take to how many people to make a first
workable version of gcc?
This is because I want (in several month) to create
a kind of compiler with a team, and I do not really
know how to estimate the time it'll take.
>>

I would say that if you don't know how to do this estimate, then
that means that you have no experience writing compilers, in which
case any estimate you get from people who do have the experience
could be misleading by a huge factor.

Re: Strange question

[Daniel Egger]

Am Son, 2002-02-03 um 23.44 schrieb David Edelsohn:

> 	It all depends on what you mean by "workable".  In Matt Dillon's
> interview on KernelTrap, he mentioned that it took him two weeks to write
> the initial version of his DICE C Compiler which could compile 'hello
> world', but many months to refine it into what one would consider a
> "commercial quality" compiler. His compiler only produced 68000 code and
> had nothing close to the optimizations in GCC.

Well, DICE C was a compiler but far away from "commercial quality". It
failed to compile anything but pretty trivial examples and lacked the
functionality to use OS headers other than those delivered with DICE
which means that I wasn't even able address all functions of the OS.

So depending on the project the new compiler should be able to handle
(and maybe even optimisation is wanted) the team should rather use
a different compiler to base their estimation on. I still have
difficulties why someone would want to start a new compiler project 
with a few man nowadays when "commercial quality" code like gcc is for
free.
 
-- 
Servus,
       Daniel

Re: Strange question

[Robert Dewar]

Typical commercial compilers represent between 10 and 100 person years
of work in aggregate. Of course a student can put together a compiler in
a semester as a course project and many have done so, so the amount of
work depends on what you are trying to do to a huge extent. Asking for
general input of this kind will likely be of absolutely no value at all.

Re: Strange question

[David Edelsohn]

>>>>> Quentin Bouvart writes:

Quentin> I'd like to ask you a strange question: how many time
Quentin> did it take to how many people to make a first
Quentin> workable version of gcc?
Quentin> This is because I want (in several month) to create
Quentin> a kind of compiler with a team, and I do not really
Quentin> know how to estimate the time it'll take.

	It all depends on what you mean by "workable".  In Matt Dillon's
interview on KernelTrap, he mentioned that it took him two weeks to write
the initial version of his DICE C Compiler which could compile 'hello
world', but many months to refine it into what one would consider a
"commercial quality" compiler.  His compiler only produced 68000 code and
had nothing close to the optimizations in GCC.

	Depending on your definition of "workable", maybe a month from a
good compiler development team.  For something commercial, many months to
a few years.  That's just a guess.

David

Strange question

[Quentin Bouvart]

Hello,

I'd like to ask you a strange question: how many time
did it take to how many people to make a first
workable version of gcc?
This is because I want (in several month) to create
a kind of compiler with a team, and I do not really
know how to estimate the time it'll take.

Sincerely

Quentin

ps: some of my sentences might be weird, that's
because i'm french

__________________________________________________
Do You Yahoo!?
Great stuff seeking new owners in Yahoo! Auctions! 
http://auctions.yahoo.com