Re: context and lilypond

[Henning Hraban Ramm]

Am 2013-01-21 um 10:05 schrieb Wim van Dommelen:

>> The snippet buffers are just numbered, so if you insert one before the 
>> first, all others will get re-rendered, too.
>> I guess we could change that behaviour, e.g. use keywords or some UID per 
>> buffer.
>> I’ll discuss that with Aditya (author of t-filter).
> 
> Could you use the md5? In one of the temporary files it is already stored, so 
> it is calculated (file xxx.tmp.md5 contains the md5 hash of xxx.tmp which is 
> a lilypond source-file, stored in the temporary directory)
> 
> If the files are not numbered (sequentially) but named according to the md5 
> hash there should be no need to re-render any already done piece of code 
> after any move. Disadvantage would be that with many changes (of a piece of 
> Lilypond code) lots of different files will float around (because a change of 
> the md5 will loose view on the original one). But that can be solved by some 
> regular cleaning.

Without changing the filter module we can’t access the MD5 before the buffer 
and the hash are written to files, and then the name is already there.

See my other mail (yesterday night): Since we can’t set a "default buffer name 
conversion"* but can set a single buffer's name, we can (and I will) just go 
that route.

> it’s as simple as adding [name=something] to \startlilypond, e.g. 
> [name=\currentcomponent]


*) Aditya told me, when I asked if I can set e.g. [name=\currentcomponent] as 
default:
> The name key is reset at every `\startexternalfilter`; otherwise it will be 
> difficult to mix `name=....` with normal usage.




Greetlings, Hraban
---
fiëé visuëlle
Henning Hraban Ramm
http://www.fiee.net
http://angerweit.tikon.ch/lieder/
https://www.cacert.org (I'm an assurer)