lilypond-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Guile - State of 2 / Release of 3


From: David Kastrup
Subject: Re: Guile - State of 2 / Release of 3
Date: Mon, 15 Apr 2019 07:29:14 +0200
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/27.0.50 (gnu/linux)

Thomas Morley <address@hidden> writes:

> Hi Jahrme,
>
> great you're interested in the topic.
> I'll try to give you some feedback to your questions.
>
> ad 1.
>
> If you do on master
> confirgure --enable-guile2
> and apply the attached patches (some are in master already, if in
> doubt prefer what's in the source already) you'll get a working
> lilypond.
> Working means `make´, `make doc´ will succeed and it survives the
> regtest with very minor issues, for guile-2.2.x and guile-2.9.1
> At least the last time I tested this.
>
> Disadvantges/issues are the slow-down.
> Other issues are listed here: https://ao2.it/tmp/lilypond-guile2/TODO
> Furthermore `procedure-source´ is disabled in guile-2, which is
> problematic for us.

I've grepped for it: I don't think that this should be an issue since it
just involves some diagnostic output that could be disabled without loss
of crucial functionality.

> In general we have no method to deal with .go-files.

Yup.  That's important for installed versions of LilyPond.

> And we're beaten by encoding-issues.

That's a strong word: most stuff works but the principal problem is that
Guile has no way to pass anything but Latin-1 strings through the Guile
API without recoding, and LilyPond works in UTF-8.  This is a major
performance problem since LilyPond is coded in significant parts in C++
and passes large amounts of text in and out of Guile.

> There's are likely more ...
>
> ad 2.
> Though, the slow-down is still huge. I can't confirm a substantielly
> speed up for guile-2.9.1
>
> Over the years several people worked on the topic.
> Speaking only for me, I think all low hanging fruits are done. And I
> don't have the knowledge to go deeper :(
> .go-files and encoding-thingies are the heaviest showstopper for now,
> all way out of my depth.
>
> If you want to start the above listed issues are the current TODO-list.

-- 
David Kastrup



reply via email to

[Prev in Thread] Current Thread [Next in Thread]