nb : I've finaly understood the performance/out of memory problems I
had for that last 6 months !!!!! It was due to the lurette.cov file
that was getting larger and larger.

Timothy Bourke (trigerred where the clock of some args are some other
args and when the names of variables are shared between the caller
and the callee).

The problem is in UnifyClock.f or (in evalClock.ml) ; the current
change in evalClock.ml fixes the pb in Tim's program, but I suspect
it is still buggy.

Translate fby into -> pre again, because the current implementation of fby is buggy if the fby is initialised with an input

Indeed, the initialisation of the fby was done when the soc was
created.  Hence the first fby that was translated was giving its
initial value to all others forthcoming fby !!!

In order to fix that, I've modified the type of Soc.key so that the
initial value is part of its key.

Note that currently, it does not work if the initial value is an input.

A few Array.copy were necessary when updating the memory.

Actually, it was a misunderstanding of mine (R1) ; the behavior was consistent
with the lv6 doc, but not with the v4 behavior...

Indeed, # means "at most 1 among n", not "exactly one among n" as I tought...

nb: test run on trieves, not triglav that is down, so the time is meaningless

Also, check programs that should fail using lurette instead of just
lus2lic. Indeed, some programs are only suposed to fail at runtime.

Change the main node names of prog in should_fail as it is necessary
for Lurette non-reg scheme

By redirecting the stderr of ecexe onto stdout, ecexe now terminate properly when called from lurette.

Hence, I have no more unresolved test (they are now failures) and the exec
time is now 98s (versus 217!!)

Some failures were actually due to wrong programs (mved to should_fail/semantics)

nb : the change has actually been done in lurette.

when the arity of an alias node was wrong.

  #fail: 83 -> 81

The rationale is to avoid local vars blow-up on some
examples. Indeed, the generated oracle blows-up (e.g., on left.lus)
if we execute it via v4, whereas via v6 it works fine.

This change triggers a couple of bugs that ware easy to fix
(confusion between div and slash) that I've fixed along the way.
For the others, I'll see later.

Overall it's a progress albeit
  #fail: 80 -> 83

indeed:
  #unresolved: 20 -> 12
  #passes		878 -> 883
  time: 335 -> 228

The test now runs in 150s (versus 417s)

[...] for some reasons, i cannot reproduce the 150s !!!

itself calls another node on a non-trivial clock (i.e., using a when)
was not producing correct code.

I've fixed this by performing the fix-point on nodes rather than on
equations. Indeed its more natural and efficient, and it avoid the
problem above. However, I did not really fix the problem, but just
turn around it. All tests seems to work fine though.

nb : #FAILS=81->80 (and -2 unresolved, but because I fixed the prog)

 nb : #FAILS=84->81

nb : #FAILS=128->90

(to be kind with data plotters).

Actually it does not fix any test, but I feel it is a (slight)
progress, so I commit it... Well, to be fair, one objective progress
is that I can now compile the test suite with the -np option.

112 tests failed, and 127 passed !!

e.g., "uminus" instead of "-"

In order to do that, I've generalised the type of merge : now the clock
argument can be any expression. Some assert false still prevent its use,
but it should be easy to get rid of them (I'll do that latter).

In order to do that, I've generalised the type of merge : now the clock
argument can be any expression. Some assert false still prevent its use,
but it should be easy to get rid of them (I'll do that latter).

nb: thus it works with red and fill since they are exactly the same !!!

F**k! socExec.ml was not gitted!!!

1) At the Lic level, there's no reason to distinguish betwenn node
calls, and predef node calls. Indeed it makes things simpler and more
homogeneous afterwards.

2) int strings are only converted when necessary (constant evaluation).

3) const are handled directly under Lic.by_pos_op instead of being under
   PREDEF_CALL, which make things easier and more logical.

Can be tried unsing the -exec option.

currently only working with trivial programs manipulating equalities
(wires) and predef operators (no pre, ->, arrays, and node calls).

Indeed, the data structure in output of the lis2loc parser was very
similar to Lic.t/LicPrg.t. Hence, this shunt.

At this stage it compiles, but it's untested and does not work.

Indeed, part of licDump was done in lic, but that part should
not depend in the compil option.

Pascal prepared « the field » for that change.