Dave Cooper (@dcooper) on GitLab, 2018-10-21.
This should be fixed by MR !68.
This Issue can be closed when !68 is merged, passes pipeline, and passes locally for other implementations people are using.
Erik Huelsmann (@ehuelsmann) on GitLab, 2018-10-21.
@dcooper I think this demonstrates a big issue with the Common Lisp ecosystem (as it's the second issue we're running into now -- the other one being aserve not loading on the SBCL version on common-lisp.net). This would never happen in a Python/Ruby/Perl environment, simply because there are far fewer possible combinations between libraries and implementations. But also: Perl has had continuous integration since the early 2000s.
Anyway: I'm not sure if this is our use or that this is a bug in cl-markdown. Are you?
Dave Cooper (@dcooper) on GitLab, 2018-10-21.
@eheulesmann Yep, it's well known that there are challenges and costs to bear in order to enjoy the flexibility and diversity of our ecosystem. And it seems that some Common Lispers (including myself) have been slow to the party of continuous integration, learning how to work with others on github and gitlab etc in a cooperative spirit with Merge and Pull requests, and so on. I'm trying to learn, and many others are obviously on board with this in a big way already.
On the positive side, I think it's bordering on miraculous that the ecosystem has achieved even what it has at this juncture, with no central body or authority coordinating everything and no large-scale funding being poured into it. Just a loosely knit "community" (if you can even call it that) of roughly like-minded individuals (a few of whom have admittedly contributed, and continue to contribute, some real heroic efforts). Compare this with the millions of $$ of funding received and consumed by e.g. Python and Julia organizations. We've had none of that (yet -- imagine what we could do if we did), but we're still here (of course, I suppose the development of the ANSI CL spec consumed a lot of $$, but we're still benefiting from the long tail of that decades later).
Anyway, the more that Common-Lisp.Net and CLF can set a good example with our own developments, the better. And maybe at some point, with sufficient support and funding, CLF and CLNet can become involved in a more official capacity in terms of hosting and managing continuous integration for the whole CL ecosystem. If not us, then who?
Dave Cooper (@dcooper) on GitLab, 2018-10-21.
Note that the two issues you cite that we've run into are very different in nature: The first was some kind of internal API mismatch which is specific to one (or a few) older version(s) of one implementation; an unfortunate sequence of events and frustrating to be sure, but one which would be invisible to those running an up to date implementation.
The second issue (with cl-markdown) turns out to be one of standards non-compliance, where a piece of code has been working in the past but starts crashing with newer and more varied implementations (so it's a bit of an opposite problem as the first, since more people would see it as time goes on). I would argue that this is actually a "good" problem, though: it's an example of the ANSI spec working exactly as intended. Newer impementations are continuing to educate themselves regarding the spec and tending to increase their standards-compliance enforcement, which forces libraries to clean up their act (often, as in this case, with the help of their users). This causes a bit of temporary pain, but in the long term will bode well for the whole ecosystem.
It is indeed good to document the process of building out this site and record what happens when we bring in more libraries and expand the functionality (and @root, this Gitlab you're maintaining for us is a huge boon in support of that). Our experience may be typical of what others encounter, so the more we can record and share about challenges and how we overcome them, the better.
Erik Huelsmann (@ehuelsmann) on GitLab, 2018-10-28.
As !68 has been merged, this issue can be closed as per the original description.
Sign in to comment.