Added current AllegroServe and replaced Python test server with it. 56
opened by common-lisp.netDave Cooper (@dcooper) on GitLab, 2018-10-17.
Added a temporary snapshot of current AllegroServe into cl-site repository (we'll have to carry it here until it manages to get into Quicklisp -- I'm working on that). Changed default make serve command to publish & serve from CL/aserve instead of Python. By default, loading build.lisp will now start serving the cl-site on localhost:8008.
Keeping the python server around for now for compatibility testing, benchmarking, etc.
0 commits, 0 files changed
Administrator (@root) on GitLab, 2018-10-19.
This turns build into serve. Why do we need that? Can't we keep build and serve separate?
I mean, the pipeline wants to build, not serve. And even if we want to serve from lisp, we don't want to do that in the pipeline: we'd want to do that after a deploy.
Administrator (@root) on GitLab, 2018-10-19.
You just moved the simplest of instructions down to the end. Don't we want the barrier of entry as low as possible? Who's going to read beyond the complex instructions to find the simple ones?
Dave Cooper (@dcooper) on GitLab, 2018-10-19.
This can be considered a test serve. I think it's reasonable that part of the pipeline should be to confirm that the test server fires up without incident.
Note that the (cl-site:start-aserve) should be tolerant of the default port already being busy (it then tries subsequent ports).
The command-line make (used by the pipeline, right?) will start serving, then Lisp will immediately exit, so it's not doing more than testing that the server can start up.
Dave Cooper (@dcooper) on GitLab, 2018-10-19.
Just reverted this. But I'm still not "recommending" command-line over running a Lisp repl (let the user make that judgement).
Dave Cooper (@dcooper) on GitLab, 2018-10-19.
I hope you don't mind - maybe I shouldn't have, but I went ahead and deleted all the long comments because I think anyone reading them in the future would get a headache and they wouldn't serve any good purpose. Sorry for having been so wordy.
Anyway, I've resolved the WIP status and took the test serving out of the pipeline. README.md is updated accordingly to let folks know how to run the test server.
Hoping this can be merged now, and we'll hold off on SBCL upgrade for another day (or just keep going with the stock one, as you mentioned)...
Erik Huelsmann (@ehuelsmann) on GitLab, 2018-10-20.
Thanks!
For the next steps, though lets focus on getting the functionality we actually need to update the site's content into the site builder (one way or another).
Sign in to comment.