[ragel-users] "declared but not used" errors with v7 Go backend
Damian Gryski
dgryski at gmail.com
Wed Dec 5 12:45:30 UTC 2018
While you're poking at the Go code generation, some small tweaks:
1) The type of the static arrays will be inferred from the type on the
right-hand side of the equals sign; it doesn't need to be explicit in the
variable declaration.
2) Go doesn't fall-through case statements by default; the `break` lines
are not necessary.
diff --git a/src/host-go/rlhc-go.lm b/src/host-go/rlhc-go.lm
index 7ca37b88..41fc2560 100644
--- a/src/host-go/rlhc-go.lm
+++ b/src/host-go/rlhc-go.lm
@@ -288,7 +288,7 @@ namespace go_gen
[embedded_host(EH)]
}
case [A: static_array] {
- "var [A.ident] \[\] [type(A.type)] = \[\] "
+ "var [A.ident] = \[\] "
"[type(A.type)] { [num_list(A.num_list)] }
"var _ = [A.ident]
}
@@ -379,13 +379,11 @@ namespace go_gen
{
"case [expr( CB.expr )]:
"[stmt_list( CB._repeat_stmt )]
- "break;
}
case [DB: default_block]
{
"default:
"[stmt_list( DB._repeat_stmt )]
- "break;
}
case [CL: case_label]
{
On Wed, Dec 5, 2018 at 9:11 AM Damian Gryski <dgryski at gmail.com> wrote:
> You can prefix the underscore assignments with var also. That will work
> both at the function and global scopes.
>
> var _scanner_trans_cond_spaces [] int8 = [] int8 { ... }
> var _ = _scanner_trans_cond_space
>
> On Wed, Dec 5, 2018 at 6:31 AM Adrian Thurston <thurston at colm.net> wrote:
>
>> Hi Damian,
>>
>> I just altered the codegen so that unused arrays are not generated. This
>> eliminates the need to assign to _ for the static arrays. The need is still
>> there for static variables, for example:
>>
>> _ = scanner_first_final
>> _ = scanner_error
>> _ = scanner_en_main
>>
>> I've noticed though that the _ assignment trick isn't valid for arrays
>> that are not in a function scope, which means we can't use it right after
>> declaration. It could go into the write exec block or the write init block
>> though. To do that I'll have to create a dedicated reference statement for
>> the intermediate language since the go code generator does not understand
>> the state machine. It's just a language translation. The reference
>> statement can have no effect in other languages.
>>
>> For the time being I'll need to revert the _ assignment since it breaks
>> ragel programs where the data is not function scoped.
>>
>> Adrian
>>
>> On 2018-11-01 09:12, Damian Gryski wrote:
>>
>> This is with ragel 7.0.0.11 and colm 0.13.0.6. I was unable to get ragel
>> to build from git.
>>
>> With the example grammar at github.com/dgryski/ragel-examples/regexp1, I
>> get the following error:
>>
>> $ ragel-go sshd.rl
>> $ go build
>> # github.com/dgryski/ragel-examples/regexp1
>> ./sshd.go:11:6: _scanner_trans_cond_spaces declared and not used
>> ./sshd.go:12:6: _scanner_trans_offsets declared and not used
>> ./sshd.go:13:6: _scanner_trans_lengths declared and not used
>> ./sshd.go:14:6: _scanner_cond_keys declared and not used
>> ./sshd.go:17:6: _scanner_eof_cond_spaces declared and not used
>> ./sshd.go:18:6: _scanner_eof_cond_key_offs declared and not used
>> ./sshd.go:19:6: _scanner_eof_cond_key_lens declared and not used
>> ./sshd.go:20:6: _scanner_eof_cond_keys declared and not used
>> ./sshd.go:21:6: _scanner_nfa_targs declared and not used
>> ./sshd.go:22:6: _scanner_nfa_offsets declared and not used
>> ./sshd.go:22:6: too many errors
>>
>> The usual solution to this in autogenerated code is to assign them to
>> `_`, the underscore "blank variable". This satisfies the compiler and the
>> entire array will be eliminated as a dead-store if it ends up not being
>> used.
>>
>> Thus, each declaration for variables which may or may not be used will
>> look like:
>> var _scanner_trans_cond_spaces [] int8 = [] int8 { ... }
>> _ = _scanner_trans_cond_space
>>
>> Note also the first `[] int8` is redundant -- the compiler will infer the
>> type from the type of the array on the right hand side of the `=`.
>>
>> Fixing these issues in the Go template in the src/host-go directory and
>> rebuilding fixed these errors (although I had to copy files from the git
>> checkout for the template changes to recompile successfully).
>>
>> Damian
>>
>> _______________________________________________
>> ragel-users mailing listragel-users at colm.nethttp://www.colm.net/cgi-bin/mailman/listinfo/ragel-users
>>
>> _______________________________________________
>> ragel-users mailing list
>> ragel-users at colm.net
>> http://www.colm.net/cgi-bin/mailman/listinfo/ragel-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.colm.net/pipermail/ragel-users/attachments/20181205/ddc55816/attachment.html>
More information about the ragel-users
mailing list