Hi Ludovic,

Ludovic Courtès <ludo@gnu.org> writes:

> Mark H Weaver <mhw@netris.org> skribis:
>
>> Ludovic Courtès <ludo@gnu.org> writes:
>
> [...]
>
>>> So there are two things.  To fix the issue you reported (build output
>>> that goes through), I think we must simply turn off UTF-8 decoding from
>>> ‘process-stderr’ and leave that entirely to ‘build-event-output-port’.
>>
>> Can we assume that UTF-8 is the appropriate encoding for
>> (current-build-output-port)?  My interpretation of the Guix manual entry
>> for 'current-build-output-port' suggests that the answer should be "no".
>
> What goes to ‘current-build-output-port’ comes from builds processes.
> It’s usually UTF-8 but it can be anything, including binary garbage,
> which should be gracefully handled.
>
> That’s why ‘process-stderr’ currently uses ‘read-maybe-utf8-string’.

I agree that we should (permissively) interpret the build process output
as UTF-8, regardless of locale settings.  However, the encoding of
'current-build-output-port' is orthogonal, and I see no reason to assume
that it's UTF-8.

As 'process-stderr' is currently implemented, it makes no assumptions
about the encoding of 'current-build-output-port'.  That's because it
uses only textual I/O on it.  The end result is that the UTF-8 build
output is effectively converted into the port encoding of
'current-build-output-port', whatever it might be.  I think that's how
it should be, no?

>> Also, in your previous message you wrote:
>>
>>   The problem is the first layer of UTF-8 decoding that happens in
>>   ‘process-stderr’, in the ‘%stderr-next’ case.  We would need to
>>   disable it, but only if the build output port is
>>   ‘build-event-output-port’ (i.e., it’s capable of interpreting
>>   “multiplexed build output” correctly.)
>>
>> It sounds like you're suggesting that 'process-stderr' should look to
>> see if (current-build-output-port) is a 'build-event-output-port', and
>> in that case it should use binary I/O primitives to write raw binary
>> data to it, otherwise it should use text I/O primitives and write
>> characters to it.  Do I understand correctly?
>
> Yes.  (Actually, rather than guessing if (current-build-output-port) is
> a ‘build-event-output-port’, there could be a fluid to ask for the use
> of raw binary primitives.)
>
>> IMO, it would be cleaner to treat 'build-event-output-port' uniformly,
>> and specifically as a textual port of unknown encoding.
>
> (You mean ‘current-build-output-port’, right?)

Yes, indeed.

> I think you’re right.  I’m not yet entirely sure what the implications
> are.  There’s a couple of tests in tests/store.scm for UTF-8
> interpretation that describe behavior that I think we should preserve.

I certainly agree that we should preserve those tests.  I would go
further and add two more tests that bind 'current-build-output-port' to
a port with a non-UTF-8 encoding (e.g. UTF-16) and verify that the λ
gets converted correctly.  The test build process would output the λ as
UTF-8, but it should be written to 'current-build-output-port' as
e.g. UTF-16.

What do you think?

>> I would suggest changing 'build-event-output-port' to create an R6RS
>> custom *textual* output port, so that it wouldn't have to worry about
>> encodings at all, and it would only be given whole characters.
>> Internally, it would be doing exactly what you suggest above, but those
>> details would be encapsulated within the custom textual port.
>>
>> However, I don't think we can use Guile's current implementation of R6RS
>> custom textual output ports, which are currently built on Guile's legacy
>> soft ports, which I suspect have a similar bug with multibyte characters
>> sometimes being split (see 'soft_port_write' in vports.c).
>>
>> Having said all of this, my suggestions would ultimately entail having
>> two separate places along the stderr pipeline where 'utf8->string!'
>> would be used, and maybe that's too much until we have a more optimized
>> C implementation of it.
>
> Yeah it looks like we don’t yet have custom textual output ports that we
> could rely on, do we?
>
> I support your work to add that in Guile proper!

For now, I can offer a new implementation of custom textual output ports
built upon custom binary ports and the 'utf8->string!' that I previously
sent.  See attached.

     Thanks,
       Mark


--8<---------------cut here---------------start------------->8---
GNU Guile 2.2.4
Copyright (C) 1995-2017 Free Software Foundation, Inc.

Guile comes with ABSOLUTELY NO WARRANTY; for details type `,show w'.
This program is free software, and you are welcome to redistribute it
under certain conditions; type `,show c' for details.

Enter `,help' for help.
scheme@(guile-user)> (load "utf8-decoder.scm")
scheme@(guile-user)> (load "guile-new-custom-textual-ports.scm")
scheme@(guile-user)> (define (my-write! str start count)
                       (pk 'my-write! (substring str start (+ start count)))
                       count)
scheme@(guile-user)> (define port (make-custom-textual-output-port "test1" my-write! #f #f #f))
scheme@(guile-user)> (display "Hello λ world!" port)
scheme@(guile-user)> (force-output port)

;;; (my-write! "Hello λ world!")
scheme@(guile-user)> (string->utf8 "λ")
$2 = #vu8(206 187)
scheme@(guile-user)> (string->utf8 "Hello λ world!")
$3 = #vu8(72 101 108 108 111 32 206 187 32 119 111 114 108 100 33)
scheme@(guile-user)> (put-bytevector port #vu8(72 101 108 108 111 32 206))
scheme@(guile-user)> (force-output port)

;;; (my-write! "Hello ")
scheme@(guile-user)> (put-bytevector port #vu8(187 32 119 111 114 108 100 33))
scheme@(guile-user)> (force-output port)

;;; (my-write! "λ world!")
scheme@(guile-user)>
--8<---------------cut here---------------end--------------->8---