Nice. What kind of hardware do we need to run this? How much traffic does the server see? Sebastiano: - could you help us with the sizing of things please?
Werner: - if we have the snippets in a "one snippet per file" setup in gitlab (or something of that general flavour), what purpose does MySQL serve? - it seems to me in this scenario the backend of LSR would simply be a "cron job" that runs "git pull" once at midnight and then regenerates all the images missing from new snippets (if any)? What am I missing here? - I guess in this scenario the "contribution interface" is simply someone that approves a "pull request" on the gitlab repo... L On Mon, Jun 30, 2025 at 5:25 PM Werner LEMBERG <w...@gnu.org> wrote: > > > what can I do to help with this situation going forward? > > Spread the word that we need people who have experience with PHP and > Java, and who are willing to help us migrate LSR to another host. > > > I must admit I feel I'm only partially clear on the goals and use of > > the LSR. Maybe if folks would list the various components of its > > functionality and pair to these the current thinking in terms of > > what should be kept vs improved vs retired, we could formulate a > > plan? > > AFAICS, the LSR serves two purposes. > > 1. It is a collection of contributed showcases, presented as small > snippets, that helps users to find out what you can do with > LilyPond. > > 2. It is also used as a source for LilyPond documentation. In other > words, quite a large group of LSR snippets is directly integrated > into the NR, and some more are also accessible as a separate > document, see > > https://lilypond.org/doc/v2.24/Documentation/snippets/index.html > > Anybody can contribute; some people serve as 'LSR editors' and approve > snippets, which make them eventually publicly available. In most > cases, this approval process includes some minor editing. > > For me, the most serious drawback of the LSR is that there is no > history preserved. If an editor removes a snippet, it is lost. My > idea is to use a git repository to keep track of all changes, > including the person's user name who actually did modify the code. > > I like the current interface to contribute and/or modify snippets. > There are certainly areas where improvements are useful and/or > necessary, but in total it serves its purpose quite well IMHO. > Consequently, I suggest the following route – but please note that I'm > have no idea whether my suggested steps are actually feasible. > > 1. Set up a new machine to run the LSR code as-is. > 2. Update the components as much as possible. > 3. Provide an interface between the MySQL database code of LSR and a > git repository to push changes as soon as somebody hits the 'save' > button in the LSR interface. > > AFAICS, this is the way of the least work. If this is done, > interested people can start with reimplementing LSR if there is a > desire to do so. > > > Werner > -- Luca Fascione
