From Guile NEWS ** Macros need to be defined before their first use. It used to be that with lazy memoization, this might work: (define (foo x) (ref x)) (define-macro (ref x) x) (foo 1) => 1 But now, the body of foo' is interpreted to mean a call to the toplevel ref' function, instead of a macro expansion. The solution is to define macros before code that uses them

This shows up in LilyPond initialization as . . [/home/ian/lilypond/out/share/lilypond/current/scm/ly-syntax-constructors.scm] [/home/ian/lilypond/out/share/lilypond/current/scm/define-context-properties.scm] [/home/ian/lilypond/out/share/lilypond/current/scm/translation-functions.scm;;; note: autocompilation is et GUILE_AUTO_COMPILE=0 ;;; or pass the --no-autocompile argument to disable. ;;; compiling /home/ian/lilypond/scm/translation-functions.scm

error: Not a markup command: line compiling markup expression (#:line (#:slashed-digit figure))

Patrick has prototyped a solution which reorders the load-order and restructures the declaration of the markup commands. This tracker is to merge the code into the master repository.

Hi Ian,

As far as I can tell, my solution works fine, but it is really hacked together.

IWBN if we can reorganize the macros and procedures so that everything is still organized and coherent.

Jan has tweaked some of this stuff in commit 05514c83a49dd3dfabd1fa7774c69c3038ecde6b , Author: Jan Nieuwenhuizen Date: Wed Feb 2 11:18:28 2011 +0100

Guile-1.9 compatibility fixes.

  * remove #:use-syntax
  * rewrite curried definitions
  * change order of loading .scm files so that
    markups are defined before used in code

Ian,

Jan did move a couple of functions around, but I still ran into compile failures as a result of markup expressions being undefined.

The reordering commits from my guile branch still "work", so I'll start cleaning them up to make the diff as minimal as possible.

@pnorcks: how is this coming along? LilyPond is now not even buildable anymore with latest guile stable-2.0

If you have a diff that works with current HEAD, I'd like to experiment with that in GUB's guile 2.0 branch.

I found git://repo.or.cz/lilypond/patrick.git

Any reason for not doing this on savannah, on dev/patrick ?

Thanks for the suggestion. dev/patrick is now on savannah.

I just test-compiled LilyPond against guile commit 900a6f87bad on the stable-2.0 branch, with mixed results. The first two attempts ended in a segfault, and the third attempt succeeded.

I'll have to take a closer look... it seems that (since 2.0) Guile is more strict with regard to this whole macro issue. I didn't see this problem with the official Guile 2.0 release.

Thanks.

FWIW, I used it for schikkers list and I haven't seen any segfaults, guile stable-2.0: f80ed1be

I was almost going to suggest to push it to master.

Thanks, Jan. I don't see segfaults anymore with the snapshot from that commit.

With regard to merging the reordering stuff into master, I think it's actually hanging by a needle and thread... So, not yet. :)

I've started work on Issue 1686 to allow us to compile our scm/.scm files to scm/out/.go and have come across another "feature". It looks like we have .scm files in the list loaded by lily.scm that have dependencies on other .scm files earlier in the list (e.g. markup.scm defining markup?).

I think we need to do some janitor work so that definitions used like this are defined in lily-library.scm (which is the first file loaded in the list).

Cheers, Ian

would it be possible to just change the order of files in the makefile? Or maybe have a SCM_ARE_DEPENDENCIES= variable (which are compiled before attempting the other scm files), or something? I'm not certain that including markups in lily-library.scm simply to work around a build system issue is a good solution.

What I'm doing in T1686 is adding support for a -d compile-scheme option, which will get lily.scm to byte-compile all the files which are currently loaded in the init-scheme-files list. We need to do it this way so we can have all the Scheme definitions available which are defined in the C++ initialization code etc.

I can't see a way of picking up a build system variable like SCM_ARE_DEPENDENCIES at run-time.

However we do it, it looks like the ordering of this list needs sorting.

Ian, check out my dev/patrick branch and see if any files from scm/ still fail to compile.

I haven't tested this in a while, so something may have broken again...

Hi, I've been doing some testing in conjunction with Issue 1686. I can get all the files to load OK in the call from ly_c_init_guile (where lily.scm itself is loaded) but it then fails in the later call to lilypond-main in lily.scm.

Here's the tail of a --verbose run. At the moment it fails with both Guile 1.8 and 2.0.

Building font database...

Processing test.ly' Parsing... [/home/ian/src/lilypond/build/out/share/lilypond/current/ly/init.ly [/home/ian/src/lilypond/build/out/share/lilypond/current/ly/declarations-init.ly [/home/ian/src/lilypond/build/out/share/lilypond/current/ly/music-functions-init.ly] [/home/ian/src/lilypond/build/out/share/lilypond/current/ly/toc-init.ly /home/ian/src/lilypond/build/out/share/lilypond/current/ly/toc-init.ly:25:33: error: syntax error, unexpected MARKUP_FUNCTION, expecting SCM_IDENTIFIER or SCM_TOKEN tocTitleMarkup = \markup \huge \column { /home/ian/src/lilypond/build/out/share/lilypond/current/ly/toc-init.ly:27:4: error: unknown escaped string:\hspace'

\hspace #1

/home/ian/src/lilypond/build/out/share/lilypond/current/ly/toc-init.ly:28:2: error: syntax error, unexpected '}', expecting '='

} /home/ian/src/lilypond/build/out/share/lilypond/current/ly/toc-init.ly:29:37: error: syntax error, unexpected '{', expecting SCM_IDENTIFIER or SCM_TOKEN tocItemMarkup = \markup \fill-line { ERROR: In procedure apply: ERROR: Wrong type argument in position 1: # ian@nanny-ogg:~/src/lilypond/build$

I've put the current patch for this and T1686 on Rietveld as http://codereview.appspot.com/4760049. Any ideas for progressing with this would be appreciated, as otherwise I'm stuck at the moment.

Cheers, Ian Hulin

Disentangled load order issues from other Guile V2 issues. Running with Guile 1.8.7 now passes make and regtests.

Patch available for review

http:://codereview/appspot.com/4849054

Ian Hulin

Comment deleted

The patch needs modifying to handle upstream changes to scm/lily.scm for logging and tracing introduced by Reinhold.

I'll also have another attempt at moving the Scheme definition of markup to scm/markup-macros.scm as per Neil's suggestion in the Rietveld review.

Is there any way I can help with rebasing/updating that patch after my changes? Reinhold

Reinhold, No problems, it's merged OK. I just wanted to make sure I didn't stamp on any of your upstream changes for loglevel.

Neil, I have had another try at moving the markup definition to markup-macros.scm but it regresses the change when running when running with V2.

I'm happy with Patchset 7 going for countdown.

Ian

Ian is handling this issue.

patch here: http://www.codereview.appspot.com/4849054

Hi Graham, I would push the patch myself, but I can't get the ssh connection to Savannah to work :-{.

I followed the instructions from the CG you sent re git push access.
I've tried generating and using both dsa and rsa public keys, and entering them on the Savannah My Account/Account Configuration page, but I still can't get git pull --verbose to work. This is frustrating - grrrrr...

Pushed as commit - f41963f0b6135c02c363f8401628e85b9bc60def Thanks to Carl for facilitating this,

Cheers Ian

Thanks for commit id.)