Re: [Clo-devel] Proposed direction for common-lisp.net in 2015
Hi Erik, Thanks for the welcome. To clarify, common-lisp.net is a valuable resource already and it provides repo hosting, so they aren't necessarily incompatible, but I worry that maintaining and improving the services that are available elsewhere comes at an opportunity cost that seems all the greater due to the small size of the CL community. Of course, there's a lot of history and context I'm missing, that's just my take as a relative outsider. I joined this list a week or so ago to ask how to clone a hosted repo. I could see the gitweb but there was no clone url provided. I was about to write to the list when, on reading another thread I saw a clone url for a different project and worked out the required transform from project to git url. I'd be happy to look into how CL resources are and could be indexed. Though time is always limited, happily CL is now my day job so I hope to become a more effective user of the language this year. BTW my email to you and the list bounced from the list. The uribl service said that one of the urls in the email was rejected, but I've entered all of them into the uribl site and they are either unknown or whitelisted. I hope this message gets through. Cheers, Andy On 7 February 2015 at 09:00, Erik Huelsmann <ehuels@gmail.com> wrote:
Hi Andrew,
Thank you for responding!
On Fri, Feb 6, 2015 at 4:10 PM, Andrew Kirkpatrick <ubermonk@gmail.com> wrote:
Hi Erik,
I'm new to this list and haven't requested any further access to the services you linked to.
Welcome! Hope you find your way around soon. Your opinion as a potential user is just as important as one of long-time users: if we succeed to convert potential users into actual users, we're achieving the goal of supporting the Common Lisp community, I'd hope.
Coming from the Perl world, what I'd like to see is something like MetaCPAN https://metacpan.org
I love metacpan indeed and I like how it uses the META.yml file to generate a lot of information. I'm not sure if the ASDF files that are currently around have enough information in them to be useful in this respect. However, it would definitely seem like something to stimulate in library authors to add the "required" elements to the ASDF files.
It is essentially a nice web interface and index to a bunch of services, most of which exist independently.
A large number of Perl modules use github for repository hosting, some use bitbucket or otherwise.
Right. In Common Lisp it's no different these days.
For issue tracking, some use the official Perl RequestTracker instance, others use github or bitbucket. Testing happens both via the Perl testers network and TravisCI on github.
For issue tracking, there's no central "standard" that projects use in Common Lisp: some have a BUGS file in the root of their development tree, others use launchpad, github or trac.common-lisp.net. As for CI, there's really no CI for individual projects within the Common Lisp community at the moment, although I know Zach Beane - maintainer of quicklisp - runs daily builds for the entire quicklisp world. Those are - as far as I understand - just builds. Not test suites.
What enables this is the CPAN::Meta spec, which is reminiscent of ASDF: https://metacpan.org/pod/CPAN::Meta::Spec
Gitlab is nice enough but it can be a pain to administer (I crashed a local install recently with a not-so-unusual construct in an Org format readme file).
Ok. Thanks for the warning :-) However, currently we have a lot of git/darcs/hg/cvs/svn repositories on a host with very little structure on the DVCS side of things. Even to the extent that Marco Antoniotti seems to have been working more than a month to migrate a project of his to Git and get common-lisp.net services working with it. One of the reasons behind this proposal is to increase the value of common-lisp.net to projects as well as new-comers like you: if there's more structure, it's easier to document for the site's maintainers how things work and for project owners to contribute to multiple projects and for newcomers to find the projects they want to contribute to.
I guess my question is, what if the site just focused on the core business of being an index to a federated community of resources?
I think your idea is perfect. We don't have such an index yet; we do have similar "services which exist independently": Quicklisp, Quickdocs, cl-user.net, cliki.net and common-lisp.net to name some. I absolutely do think there's room for such a meta service in the community.
Do I understand your suggestion correctly that you see no value in hosting any kind of code or project on common-lisp.net? Over the past 10 years, common-lisp.net has been hosting Common Lisp projects, providing them more-or-less with the services they needed. Due to the number of projects being hosted on the infrastructure it also became more-or-less of an index. However, the more-or-less-index roles can also be claimed by cliki.net and cl-user.net.
While there is absolutely an opportunity to grow common-lisp.net with the meta function, I personally (in the shorter term) only have time to work on the hugely necessary improvements to be made to common-lisp.net, the current site and infrastructure itself. The board of the Common Lisp Foundation (the stewards of the common-lisp.net domain) heartily welcome efforts to support the use of Common Lisp. Do you feel that you could work on an ASDF based meta index?
Next to that, there could be other services for which we think there's room also room for CI (Continuous Integration) specifically directed at Common Lisp projects (there's cl-test-grid, which is great, but doesn't provide continuous integration for individual Common Lisp projects - rather it has been targetted so far at assessing quality of Quicklisp distributions and providing feedback thereof to library and implementation developers).
Hopefully I haven't offended anyone with these suggestions.
Definitely not offended here. I'm glad you took the time to respond.
-- Bye,
Erik.
http://efficito.com -- Hosted accounting and ERP. Robust and Flexible. No vendor lock-in.
Hi Andrew, Thanks for the welcome. To clarify, common-lisp.net is a valuable
resource already and it provides repo hosting, so they aren't necessarily incompatible, but I worry that maintaining and improving the services that are available elsewhere comes at an opportunity cost that seems all the greater due to the small size of the CL community.
I agree that repo hosting and issue tracking can be found elsewhere. However, currently we are already providing the service. My point being that I think that - even though there's an initial "time investment" to it - the proposed direction will make it much easier to do the job of the maintainers, since there are quite a number fewer components to be configured, maintained and updated.
From the above, I conclude that you're fine with the proposed role for common-lisp.net and would even settle for less, if less relates to services available elsewhere. Looking at it that way, you support the proposal :-) Taking your advice, I'll make sure to be careful not to spend too much time on the migration in areas where such effort wouldn't be justified by efficiency on part of the maintainers or usability for end-users.
Of course, there's a lot of history and context I'm missing, that's just my take as a relative outsider.
Yes. And because you're an outsider, you have a much clearer view on our situation than I do :-) [troubled with all the history of the past 12 years of common-lisp.net hosting...]
I joined this list a week or so ago to ask how to clone a hosted repo. I could see the gitweb but there was no clone url provided. I was about to write to the list when, on reading another thread I saw a clone url for a different project and worked out the required transform from project to git url.
I'd be happy to look into how CL resources are and could be indexed. Though time is always limited, happily CL is now my day job so I hope to become a more effective user of the language this year.
Well, maybe one or two people could join to lift the weight together?
BTW my email to you and the list bounced from the list. The uribl service said that one of the urls in the email was rejected, but I've entered all of them into the uribl site and they are either unknown or whitelisted. I hope this message gets through.
Thanks for the heads-up. The error has been fixed in the mean time. Turns out the URL wasn't blacklisted, but the blacklist doesn't service entities going through Google's DNS or OpenDNS (which we're using with our hosting service). [I've implemented a way to do the blacklist lookups directly at the RBL's DNS servers.] -- Bye, Erik. http://efficito.com -- Hosted accounting and ERP. Robust and Flexible. No vendor lock-in.
participants (2)
-
Andrew Kirkpatrick -
Erik Huelsmann