Re: problematic commit 02fe038744e634b42f1a3377c4f0dc3d25e80344

[Neil Puttock]

On 24 October 2010 13:17, Valentin Villenave <address@hidden> wrote:
> On Sun, Oct 24, 2010 at 1:40 PM, Neil Puttock <address@hidden> wrote:
>> On 24 October 2010 12:32, Valentin Villenave <address@hidden> wrote:
>>> Feel free to revert it.
>>
>> TBH, I'm sorely tempted to myself.
>
> Then please do.
>
>> There's a memory leak from `make-void-music', which suggests you
>> haven't done a regression test check or run make doc.
>
> I have not, since the new function isn't used anywhere.
> I didn't suspect a stupid music-function hack that is not invoked
> anywhere could cause such things.

Well, to be fair, it's not the music function that's the problem;
rather, you've neglected to make `make-void-music' a thunk, which
means it's evaluated immediately music-functions.scm is loaded.

>> I don't think it's appropriate to gulp files and evaluate code for
>> this.
>
> I never said it was.

Sure, I understand that (your comments in the music function suggest as much).

> The only point was to make the patch as unnoticeable as possible, in
> order to get it accepted.

Believe me, I understand your frustration at waiting for GLISS to
start, but a change like this needs a more considered approach which I
don't think we can afford at the moment.

> OK. Reverting.

You can keep the `make-void-music' if you fix the bug (it just needs
parentheses around the name).

Cheers,
Neil