Verilog: pass instance parameters by name everywhere; remove -v95

[nanavati]

Instance parameters are now always passed by name; the Either in vi_inst_params is gone, along with the -v95 flag that was its main client. Follows the #997 discussion; supersedes #1021 (if this is preferred, #1021 can close — if #1021 lands first, this rebases trivially).

The insight

The one non--v95 client of positional parameters was the Classic foreign-function-bound-to-a-module flavor (in-tree: Fork.bs), where the hand-written module's parameter names were "unknowable." But the declaration does name each parameter slot: the type variable. So the convention becomes: the module's parameter names are the declaration's type variable names, and bsc emits

Fork #(.iw(32), .ow(64)) instance_vfork_0(.i(...), .o(...));

Fork.bs conforms to the existing Fork.v by renaming its own type variables (foreign vfork :: Bit iw -> Bit ow = "Fork",("i","o")) — no Verilog changes. This is a strict upgrade in failure mode: the old positional contract (type-argument order must match the .v parameter declaration order) failed silently as a miswired netlist; a name mismatch fails loudly at Verilog elaboration.

What's in it

• Name threading: MakeSymTab captures the declaration's type variable names in quantification order (the same order the type arguments arrive at applications) into VarForg; they ride ICForeign.fTyVarNames and pair with the concrete values in AConv, so ANoInlineFun now carries [(String, Integer)].
• Width discipline (new error T0159): for a foreign function bound to a module, every bit width must be a bare type variable or a literal. A type-function width (Bit (TAdd n 1)) has no name to pass and would force the .v to reimplement bsc's type arithmetic (with its exact ceiling/rounding semantics) to size its ports — rejected, with guidance to give the derived width its own variable. "True" foreign functions (no port list) are unaffected, as are BSV (* noinline *) functions (enforced monomorphic by T0111, hence parameterless).
• Either collapse: vi_inst_params :: [(VId, Maybe VExpr)], matching how ports have always worked; printer simplified; pv95params deleted.
• -v95 removal: the flag's only remaining substance was positional parameters plus dropping $signed/$unsigned (replacing the cast with a comment — lossy, not compatibility). It also hasn't delivered V95-clean output in years: the shipped primitive library uses localparam (SizedFIFO, Probe) and named parameter instantiation (SyncFIFOLevel) ungated. Relands the flag-removal intent of chore(comp): remove -v95 flag #719/Slightly modernize Verilog, take 2 #400 (cc @thoughtpolice); $signed is now always emitted. User guide updated.
• Format bumps: .bo (ICForeign gains the names field) and .ba (ANoInlineFun shape) headers bumped to 20260706-1.

Migration

For out-of-tree Classic code using foreign ... = "mod",(ports) with numeric type arguments (believed to be nearly nil): name the declaration's type variables after the module's parameter names (or vice versa). Everything else is unaffected.

Testing

• bsc.lib/fork: passes, plus a new find_regexp pinning the named emission (.iw(...), .ow(...)) in the generated Verilog.
• bsc.verilog/noinline: passes, plus a new negative test (BadForeignWidth.bs → T0159).
• bsc.verilog/v95 → renamed bsc.verilog/verilog_params; keeps testing parameter ordering and $signed output without the flag.
• bsc.options: help/print-flags goldens updated.
• Full testsuite: 18,345 PASS / 0 unexpected FAIL / 129 XFAIL / 0 XPASS (fullparallel). Correction: the long tests were silently dropped from this particular run by the parallel-make race that testsuite: fix parallel-make race that silently drops long tests #1006 fixes. A later run on this branch's continuation (Permit numeric provisos on foreign functions, checked then erased #1023) with long tests enabled shows the only long-test failures are pre-existing on stock main and unrelated to these changes (the MPEG4 output diff that testsuite: fix MPEG4 (sort ordering, check_vcd arg, bad bitstream sentinel) #995 fixes, and a stack overflow in log2_loop, cf. Fix quadratic blowups and stack overflows #957) — verified by reproducing both on an unmodified main-based build.

Related: #997, #1021, #719, #400.

🤖 Generated with Claude Code

[matx-amy]

reviewed, 👍

[nanavati]

Rebased onto current main (941eecf) and force-pushed (cf7dc26). What the rebase resolved: Re-applied named-parameter emission across upstream's port-splitting part 2: foreign-call VMInst port construction is now grouped per argument (iports ++ oports, instance list form) — plus the GenABin header refactor (headerBS).