Menu
▾
▴
#1199 fast lilypond processing via localhost network server
Originally created by: *anonymous
Originally created by: percival.music.ca@gmail.com
About 5 years ago, Han-Wen created a "proof of concept" lilypond server. It accepted a telnet (or netcat) connection, generated output for the given lilypond file sent via telnet, etc.
It would be nice if:
- somebody found that file from the mailing list archives
- figured out how it worked
- cleaned it up a bit (maybe?)
- added it to the docs
Discussion
Originally posted by: wbsoft
Ha, finally found it in the archives.
http://www.mail-archive.com/lilypond-devel@gnu.org/msg08415.html
Originally posted by: percival.music.ca@gmail.com
More important than Postponed.
Labels: -Priority-Postponed Priority-Low
Originally posted by: PhilEHol...@googlemail.com
Not convinced anyone would telnet to a server these days to compile a lilypond file. Doesn't http://weblily.net/ effectively do the same thing?
Originally posted by: percival.music.ca@gmail.com
This isn't about telnet; it's about having a "persistent" lilypond with guile loaded in memory. This eliminates about 0.5 seconds from every time you run lilypond. It just so happens that this implementation uses netcat and telnet, rather than local pipes or sockets or all the other ways of inter-process communication.
Now, if you're compiling a big score that takes 30 seconds, then saving 0.5 seconds isn't significant. But if you're compiling 200 one-line files that each take 0.75 seconds, then saving 0.5 is a *huge* benefit.
Originally posted by: bealings...@googlemail.com
It's a bit difficult to tell that it's not about telnet, given that this is the title of the issue and persistent loading of guile hasn't been mentioned until I queried the issue. Personally, I don't see saving a few 10s of seconds processing 100 files as worth any effort at all, but there you go.
Originally posted by: PhilEHol...@googlemail.com
By the way - that last comment was me - signed in with a different username on a different browser.
Originally posted by: percival.music.ca@gmail.com
... ok, I've changed the title.
The speed of processing can make a huge difference. When generating 5000 files which normally take 0.75 seconds each, saving 0.5 seconds on each file will make the entire job take 20 minutes instead of 62 minutes. And yes, I've generated that many files as part of my Masters thesis on computer-generated sight-reading exercises.
In some situations, it's even more important. Back in 2006, I was doing interactive sheet music -- I generated a short .ly file in concert, displayed one line of music to the performers, then generated a second line of music in response to my real-time actions while the performers played the first line. The latency between me finishing "acting" a line of music and displaying it for the performers could make a huge difference in how difficult this sight-reading was -- good musicians do not read a score linearly; their eyes flick back and forth constantly. Being able to see the general shape of music to come 8 beats in advance, instead of 2 beats in advance, makes the "live sight reading" much more palatable.
Summary: fast lilypond processing via localhost network server
Originally posted by: pkx1...@gmail.com
(No comment was entered for this change.)
Labels: -Priority-Low