Enforcing assumptions

[Robert Goldman]

On 9/23/16 Sep 23 -12:41 PM, Faré wrote:
> On Sep 23, 2016 12:25 PM, "Robert Goldman" <rpgoldman at sift.net
> <mailto:rpgoldman at sift.net>> wrote:
>>
>> On 9/23/16 Sep 23 -9:52 AM, Faré wrote:
>> > 1- The requirement to always use make-operation currently only applies
>> > to code within ASDF itself and well-behaved extensions. Before you
>> > enforce it more widely, you have to make sure no one in Quicklisp does
>> > it. You could have shared initialize check that: a) no instance of that
>> > class with given initargs exists yet in *operations* and hopefully b)
>> > there initargs are always null (goodbye, make-build!)
>>
>> Your docs clearly say that these MUST be built this way.  If you want to
>> say "must" then I will enforce "must."  If you want "may," we can have
>> "may" as in "For more efficient functioning, please use MAKE-OPERATION
>> to create all operation instances."
>>
> Well, at this moment it's only a performance optimization. The goal of
> enforcing the invariant would be to eventually support operations with
> meaningful initargs. So the question is: where are you going and/or what
> options do you want to leave open? Enabling initargs? Forbidding them?
> Either way, make-operation can help you enforce the invariant. And
> either way requires more work. But using make-operation makes the code
> more future-proof.
> 
>> Please articulate why you say "All operation instances MUST be created
>> through this function"?  I like the idea of having a clean API, but I
>> also note that there is code that effectively interns the results of
>> MAKE-OPERATION. What is the design rationale for this?  Do you wish to
>> be able to EQ-compare them?  If not, why do you intern these objects in
>> the *OPERATIONS* table?  Is it really so expensive to create an
>> OPERATION that we need to memoize?  Or is it because we simply create so
>> many, and this improves GC behavior?
>>
> Better GC is nice, but yes, having tables indexed by a pair of operation
> and component would be much better than having hacks like node-for...
> and it requires everyone using make-operation.
> 
>> WRT Quicklisp:  I do NOT accept that I have to check all of Quicklisp
>> for ANYTHING. I am willing to be INFORMED by results of such checks, but
>> I do NOT have time to make such checks.
>>
>> I don't expect other people to test my SIFT code, and I don't have time
>> to check arbitrary CL libraries, much less arbitrary CL libraries on
>> arbitrary implementations and operating systems.
>>
>> So, no.  If someone else wants to test other things.
>>
> You surprise me. Historically, you've always been a strong voice for
> backward compatibility and being extra cautious about not breaking other
> people's code, especially not so without extra warnings and heads up,
> even more so if we're reversing something the manual used to advertise.

I'm not saying that I don't want to avoid backward compatiblity issues.
I just think it's too much to say "you can't change anything unless you
check some large N of open source libraries by some small M of lisp
implementations by some smaller P of operating systems."  That's clearly
impossible.

One thing about having pre-released ASDF versions is that we can run
things up the flagpole and see how much we break.

But back at you -- you say you'd like to have MAKE-OPERATION be the
pinch point, so that we can check initargs, exploit the fact that
operations are unique and EQ-checkable, etc.  But then you tell me I
can't enforce the use of MAKE-OPERATION.

I don't get it.  I don't see how you can have it both ways.  The
alternative seems to be "we're going to let you write and run code that
we KNOW will break, because we don't want to signal errors."  Having the
*appearance* of backwards-compatibility is worse than obviously breaking
backwards-compatibility.

So, please: either let's get rid of MAKE-OPERATION, kill its memoizing,
and go back to letting people use MAKE-INSTANCE, or let me get on with
enforcing its use.

I'm happy to see MAKE-OPERATION die (or be left around for backwards
compatibility as just a shell around MAKE-INSTANCE), and reduce the code
size and complexity of ASDF.  Or I'm happy to enforce its use.  But the
midpoint is untenable.

I guess the final point is: please don't use the word "must," unless you
mean "must."  I feel like you're stabbing me in the back when I've tried
to take your stricture seriously, and suddenly it turns out "must means
may."
> 
> I wouldn't consult Quicklisp when implementing a big fix or making a
> backward compatible change. But when proposing a backward incompatible
> change, I make sure to warn all users that I can find.
> 
>> > 2- No, there was never a requirement that defsystem should only be used
>> > within a .asd. Actually, the test system relies heavily on the opposite.
>> > The requirement is that .asd files be loaded in the correct context, by
>> > load-asd -- notably, the correct *package* must be bound, the correct
>> > readtable, etc.
>>
>> Unfortunately, DEFSYSTEM is the only entry point we can check.  So if
>> LOAD-ASD is important, that's the only place I can check it. What else
>> would you have me do?  Check *load-truename* for "asd"?
>>
>> Look, if you want to push this, then you can't object to my enforcing
>> it.  If you don't want to push this, then we should make sure DEFSYSTEM
>> works outside the context.  But I don't want to field bug reports where
>> someone says "I tried to define this system and it didn't work," and
>>
>> In the olden days, we relied on programmers to make sure that the
>> context for DEFSYSTEM reading was appropriate.  That was a pain for them
>> sometimes, but it was clear, and it kept ASDF simple.
>>
>> At some point, ASDF decided to take on the burden of establishing the
>> context for DEFSYSTEM reading.  OK, not my choice, but a reasonable
>> decision.  But I flatly refuse to maintain *both* the DWIMing in
>> LOAD-ASD *and* DEFSYSTEM execution in arbitrary contexts.  And as a
>> programmer, I don't want ASDF to let me evaluate a DEFSYSTEM form only
>> to beat me up because some invisible context, only apparent through
>> reading the code, means that it doesn't work.
>>
>> Pick one, DWIMing, or freestanding execution, but you get only one.
>>
> There too I don't understand your reasoning. Who are you helping, and
> who are you harming?
> 
> There always was a constraint on loading a .asd, and never was one on
> using defsystem. We can assume that all software that works fits those
> constraints, but might not fit tighter ones. Why make it harder to use
> defsystem? What would you check exactly?

I don't get it.  You are saying on the one hand "you have to have the
right context to evaluate the contents of an asd file," and on the other
hand "don't stop me from evaluating ... the contents of the asd file."
What is it about an ASD file reading that requires this context, if it's
not the

Loading an ASD is not a lisp thing, though.  Loading an ASD file
involves evaluating the forms inside the file.
> 
> To me, setting the syntax context is an essential service to provide to
> whoever writes a .asd file. The .asd maintainer cannot assume anything
> about the syntax context used by the user, who may not control the
> syntax context from the end-user. If you can't even trust the package or
> readtable, or character encoding, you can't even write code that has any
> guaranteed meaning. Note that it's called .asd rather than .lisp for a
> good reason: it's Lisp code, but supposed to run in a specific context.

But that's true of EVERY Lisp file!  If you just write arbitrary code,
that relies on symbols from specific packages, and syntax from specific
readtables, and you don't put an IN-PACKAGE and readtable specification,
it will break.  But I don't try to control what you do when you are
writing arbitrary lisp code.  So clearly it's not "essential." Why do we
do it for ASD files?  What's so hard about leaving it to the programmer
to get this right?  Why did we have to make that our job instead of the
programmer's job?  Worse, we have broken that invaluable debugging tool
of interactive execution of code, by encouraging programmers to rely on
our use of LOAD-ASD.

Actually, when I read your reasoning, I think I see that this *is*
generally true (especially your remarks about the readtable). This
suggests to me that what you are really doing here is chewing off a tiny
corner of CL, and in that tiny corner, you are trying to fix a thing
that is broken about the language -- but only in that tiny corner.

I concede that this ship has sailed, and there's no tearing down
LOAD-ASD, although I dearly wish I could.

But again, I feel very frustrated, because I feel like you are saying to
me "there's SOMETHING in every .asd file that should only be evaluated
in the context of LOAD-ASD, but I'm telling you it's not DEFSYSTEM, and
I'm not telling you what it is."  Or even, "asd files should never be
loaded, or evaluated, except through LOAD-ASD, but any form in an asd
file can be."

So, ok, what IS it about an ASD file, that is not DEFSYSTEM, that must
be evaluated inside LOAD-ASD? If it's just "you could wreck the syntax
behind my back," yeah, that's true, but you could wreck the syntax
behind my back and wreck the library I'm loading, too, so why is it
ASDF's job to fix that here?  If LOAD is broken, why is it our job to
make LOAD-ASD and fix a tiny corner of the problem?


> 
> I don't see who is served, or what feature is enabled, by introducing
> this backward incompatibility.
> 
>> > On Fri, Sep 23, 2016, 09:49 Robert Goldman <rpgoldman at sift.net
> <mailto:rpgoldman at sift.net>
>> > <mailto:rpgoldman at sift.net <mailto:rpgoldman at sift.net>>> wrote:
>> >
>> >     I have been working on enforcing assumptions recently added to
> the ASDF
>> >     docs.  Specifically, that OPERATION instances only be created by
>> >     MAKE-OPERATION and SYSTEMs only be parsed inside LOAD-ASD.
>> >
>> >     I should have a merge request up for review soon, but find that it's
>> >     more tricky than I expected, because we don't even play by the rules
>> >     ourselves!  Specifically, there are calls to MAKE-INSTANCE on
> OPERATIONs
>> >     in the ASDF codebase itself.
>> >
>> >     Cheers,
>> >     r
>> >
>>
>