Closed gregcman closed 4 years ago
would you like to register?
Yes, I would register it. But each ".asd" file needs :author
, :description
, and :license
.
:author "SANO Masatoshi, cxxxr"
:license "MIT"
:description "????? system for lem: the editor for Common Lispers"
Is this good?
Also, some ".asd" files say ;;don't edit
like lem-mouse-sgr1006. Is this from roswell/project?
ros project author -d/SANO Masatoshi, cxxxr
ros project license -d/MIT
And not implemented:
ros project description -d/?????
???
ros project is not perfect and "don't edit" is to avoid breaking file by hand editing... actually you can edit. I feel I'm the second contributor and I don't need to be written in.
@cxxxr 何が質問されているかはわかると思うんだけど、どう思う?
@pupcraft I'll edit at least some of asd's author/description.
@snmsts quicklispへの登録はやってもいいと思います。 authorはシステム毎に一番変更量が多い人の名前が良い気がしますが ディレクトリ毎移動しているのでそこは注意したほうがいいですね。
translation by snmsts I think registration for quicklisp might be ok. The "author" written there could be the person who wrote the most in the system. It would be better to notice/consider directories have changed.
@cxxxr your name in author field still "cxxxr"?
@snmsts yes
https://github.com/cxxxr/lem/commit/8510a9a6f7dbf90b56761d34a35210f8e3aa5f9c please check @nz-tcoder @Hamayama @fukamachi @gos-k @gosuke-yoshita @stacksmith @t-sin @cxxxr @myaosato
I'm OK. Thank you!
The "author" written there could be the person who wrote the most in the system.
You can list all authors there, no problem.
I see that the ASD systems at https://github.com/cxxxr/lem/blob/master/lem.asd still do not have the required information.
@phoe Thank you.
You can list all authors there, no problem.
I believe we know...
Thanks for credit! I think it's OK.
simple description todo.
@snmsts do you help adding descriptions? It seems like a lot work and I can try to help if you want
@pupcraft Yes. Any help would be welcomed.
quicklisp registration todo.
Actually I change my mind, Lem doesn't have to be on quicklisp, sorry for the trouble @snmsts
Registering on quicklisp is hard
I'll just use https://github.com/snmsts/github-install or roswell
You can delete the edit-asd
branch
Thank you for helping!! And sorry for underestimating the difficulties of quicklisp and wasting all you guys's time...
I think it would nonetheless be useful for the Lisp community in general if lem is available on Quicklisp - QL is much more popular than github-install.
If you don't have the time/energy for doing this and if need any support with making lem available, please let us know - I think there are people who can support you with all the needed work.
what happend @pupcraft ?
Why put lem on quicklisp?
I don't know...sometime I feel software which are not on ql is not exist for common lisp society. publicity for me?
Registering on quicklisp is hard
What exactly is? I don't understand the issue. All ASD systems need to have basic information filled into them, and the package has to build without errors - these are the only two real requirements.
@phoe registering on quicklisp takes a relatively long time, and requires coordination. So maybe the best way to distribute software is to start with github-install or qlot or roswell first, and switch when quicklisp updates.
I would like to see this happening. How about you trying at least submiting to http://ultralisp.org? Which is a descentralized, not peer-reviewed, repository for quicklisp. And for loading from any machine you can use quicksys
@ryukinix here it is pending for ultralisp: https://ultralisp.org/projects/cxxxr/lem
quicksys looks pretty good. I wonder how the lisp ecosystem will evolve
sorry, accidentally closed and reopened
Good to see that! @pupcraft
quicksys looks pretty good. I wonder how the lisp ecosystem will evolve
Nice to hear it, I'm waiting Xach approval to be on quicklisp, they said to me that the project it's ok but seems that quicklisp is not available in the environment of project building... which is a problem for a project like quicksys. You can find more info here.
https://ultralisp.org/projects/cxxxr/lem is now live!
@ryukinix sounds like a circular problem, having to build quicklisp to be on quicklisp but quicklisp not being available. Do you think ultralisp is the best dist to use in the meantime?
unfortunately ultralisp does not seem to recognize nested asd files. so https://ultralisp.org/projects/cxxxr/lem is not live...
I'm doing my best to make Ultralisp is the best place for hosting Common Lisp libraries, guys!
And any help is very appreciated ;-)
The biggest problem right now is that site sometimes hangs because it uses lfarm to delegate checking of the new updated into a separate docker container, and lfarm seems does not close sockets when it loses connection to a worker. And during the time, server eats a limit of available file descriptors and hangs.
However, this problem does not affect already built distribution, because they are served directly from Amazon S3.
and lfarm seems does not close sockets when it loses connection to a worker.
I do not see an open bugticket on https://github.com/lmj/lfarm/issues related to this. Have you made one?
@pupcraft you are correct about nested asd files. Their support was turned off intentionally, because some libraries have broken or test nested asd files. Something like skeletons with template placeholders, or example systems. That is why I decided to not parse them by default.
But I want to add support for an explicit list of asd files as a per-project setting. Feel free to create such an issue and help with development. Or just wait while somebody else will implement this feature.
@phoe I didn't create an issue on lfarm yet, because this is just a hypothesis which has to be checked first.
I just copied lem directory in local-projects/
. It loads and builds nicely. I don't understand why can't Quicklisp just pack it. I tried Roswell and didn't like it, I honestly prefer Quicklisp only.
That's problems of Common Lisp package distribution it's a quite bad. There is a giant efforts of Xach and ASDF authors (that's guy are heros!), but I hope one day we can kill all that problems as @svetlyak40wt are trying a really nice effort to bring to us a less painful way to publish/deliver/control dependencies and projects.
We need really to unify forces for the greater good.
@ryukinix how do you think we should unify forces?
Hi @pupcraft, I'm glad that you ask.
In my view, i think we need freedom to distribute software without peer-review, as first problem to solve. That's the way modern software development works. That thing of a centralized Xach-review approval it's not a good thing.
The second problem it's make it easy to install, which Roswell are making a good job on this.
About Quicklisp repository, I prefer the more liberal @svetlyak40wt approach by giving the freedom of user to publish your own software with fast releases when he wants without need of approval.
The way we can unify forces I would suggest something like help Alexander to make ultralisp better and more easy to install greatly. Maybe a automation script ignoring the steps to deal with quicklisp stuff and just use as lib? Something like that.
Quicklisp it's great as tool, the main problem and it's publishing software as we see here. Unfortunately ultralisp have your own problems too and for that reason we still cannot publish lem
on there (yet).
But we, as Common Lisp Community, can do a great thing.
In my view projects like Roswell and Ultralisp are awesome ideas and deserves more attention.
In my view projects like Roswell and Ultralisp are awesome ideas and deserves more attention.
Totally agree. I used to use clbuild before Quicklisp was available and I miss the ability to simply add repos to my system list. What I didn't like about Roswell is it's somewhat invasive. As I used it, it downloaded SBCL and Quicklisp, even though I already had both in my system. It should, by default, look for or at least ask. Running 'ros -h' didn't tell me how to do that. I tried ros as-is anyway, but as I tried to copy my quicklisp into .ros, it broke. I didn't find anything about using my own Quicklisp and SBCL with Roswell, then I gave up.
The system manager needs to adapt itself to what I already have installed, not the other way around. For newbies, on the other hand, or people who are willing to reconfigure their on environments, Roswell seems very good.
I don't understand why can't Quicklisp just pack it.
@jessymilare Seems like https://github.com/quicklisp/quicklisp-projects/issues/1624 has an answer - Quicklisp cannot build projects that themselves depend on Quicklisp, and it seems that lem has such dependencies somewhere.
Invasive...sad to hear. I support the idea to install lem without roswell. But could you stop negative comment on my product? Do you need to say that? for what purpose? I think "it would be nice if I can install lem without roswell" is enough.
thanks.
I had no intention to offend anyone. I apologize if it sounded like that. My only purpose was to point to one particular limitation of Roswell IMO. You might agree with me or not, change it or not, that is up to you.
I don't understand why can't Quicklisp just pack it.
@jessymilare Seems like quicklisp/quicklisp-projects#1624 has an answer - Quicklisp cannot build projects that themselves depend on Quicklisp, and it seems that lem has such dependencies somewhere.
That is worse. I think we can't have hope on that, the first issue about that has 7 years. https://github.com/quicklisp/quicklisp-projects/issues/425 I proposed quicksys as I said earlier to make easier to deal with another distributions, but 7 years... I don't know if Xach really cares about this specific problem enough or it's a so hard/core problem that need to re-write the whole quicklisp (?)
I don't understand why can't Quicklisp just pack it.
@jessymilare Seems like quicklisp/quicklisp-projects#1624 has an answer - Quicklisp cannot build projects that themselves depend on Quicklisp, and it seems that lem has such dependencies somewhere.
That is worse. I think we can't have hope on that, the first issue about that has 7 years. quicklisp/quicklisp-projects#425 I proposed quicksys as I said earlier to make easier to deal with another distributions, but 7 years... I don't know if Xach really cares about this specific problem enough or it's a so hard/core problem that need to re-write the whole quicklisp (?)
It shouldn't be hard to remove the dependency on quicklisp
and dynamically test for its presence at run time. That should be better in any case. fgrep -ri "quicklisp"
shows only 19 occurrences in source code.
Also grep for ql
- the nickname of the quicklisp
package.
But yes, depending on everything at runtime should be possible - it is possible to use uiop:symbol-call
instead, for example.
Okay, first attempt. Comments are appreciated ^_^
Thank you for making the PR! Commented on the code.
OK - Xach was able to perform compilation. http://report.quicklisp.org/2019-06-25/failure-report/lem.html are the errors that lem has generated as it was being built.
OK - Xach was able to perform compilation. http://report.quicklisp.org/2019-06-25/failure-report/lem.html are the errors that lem has generated as it was being built.
About lem-xcb, I was told it wasn't supported anymore. If that is the case, maybe it's time to get rid of it. I intend to create a LTK frontend, which I believe will be more stable and maintainable, not sure when I'll have time, though.
I fixed a few typos and system dependencies. Many of the failures happend because the system definition lacks dependency on lem-core
. However, there is a "don't edit" comment above the defsystem
form of these systems, so I guess they are auto-generated and I didn't edit them. I might check that later.
I'll drop xcb. @jessymilare which system would you like to edit?
done 4170ff918f99fbd13878de7702a4e25e183a58c2
Can we have lem in quicklisp?