info
Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Teacup and Teapot
219 views
Skip to first unread message
tclwa...@gmail.com
Jan 9, 2018, 12:10:49 AM1/9/18
to
Are Teacup and Teapot being considered seriously by the Tcl community and core team
I notice that there is almost no development being done on them on the github repository
https://github.com/ActiveState/teapot
Are there any plans to maintain this and make it CPAN for Tcl
I notice that there is almost no development being done on them on the github repository
https://github.com/ActiveState/teapot
Are there any plans to maintain this and make it CPAN for Tcl
jsunth...@gmail.com
Jan 13, 2018, 7:34:24 AM1/13/18
to
Gerald Lester
Jan 13, 2018, 1:49:34 PM1/13/18
to
him an email.
--
+----------------------------------------------------------------------+
| Gerald W. Lester, President, KNG Consulting LLC |
| Email: Gerald...@kng-consulting.net |
+----------------------------------------------------------------------+
Ali M
Jan 16, 2018, 12:32:23 AM1/16/18
to
On Saturday, January 13, 2018 at 1:49:34 PM UTC-5, Gerald Lester wrote:
> On 01/13/2018 06:34 AM, j...@gmail.com wrote:
> > On Tuesday, January 9, 2018 at 12:10:49 AM UTC-5, Ali M wrote:
> >> Are Teacup and Teapot being considered seriously by the Tcl community and core team
> >>
> >> I notice that there is almost no development being done on them on the github repository
> >>
> >> https://github.com/ActiveState/teapot
> >>
> >> Are there any plans to maintain this and make it CPAN for Tcl
> >
> > In the #tcl irc channel it is derided as abandoned. A couple days ago I got the idea for a distributed software repository based on fossil, a "parent" instance to be hosted at http://dobee.tk. I'll try to remember to update this when I've got more put together. Notable first entries: Tcl 8.6.8 Tclkits for Linux x64 and ARM, A fossil compiled with tcl-integration enabled, the latest Gnocl. I'm hoping to script out to grab all the software in the Teapot and put them in the Tcler's Dobee, though probably just the latest version of each to begin with.
> >
>
> Joe Mistachkin has already built such a system -- you might want to drop
> him an email.
>
having a CPAN for Tcl , will really be great, hopefully something will be done by the time for tcl 9
> > On Tuesday, January 9, 2018 at 12:10:49 AM UTC-5, Ali M wrote:
> >> Are Teacup and Teapot being considered seriously by the Tcl community and core team
> >>
> >> I notice that there is almost no development being done on them on the github repository
> >>
> >> https://github.com/ActiveState/teapot
> >>
> >> Are there any plans to maintain this and make it CPAN for Tcl
> >
> > In the #tcl irc channel it is derided as abandoned. A couple days ago I got the idea for a distributed software repository based on fossil, a "parent" instance to be hosted at http://dobee.tk. I'll try to remember to update this when I've got more put together. Notable first entries: Tcl 8.6.8 Tclkits for Linux x64 and ARM, A fossil compiled with tcl-integration enabled, the latest Gnocl. I'm hoping to script out to grab all the software in the Teapot and put them in the Tcler's Dobee, though probably just the latest version of each to begin with.
> >
>
> Joe Mistachkin has already built such a system -- you might want to drop
> him an email.
>
Keith Nash
Jan 16, 2018, 10:50:36 PM1/16/18
to
How much development do Teacup/Teapot need? They are mature code,
sufficiently robust for ActiveState and others to run on public servers.
When they were first released, it took a little while to get used to them;
but they quickly became indispensable. Now that they are open-source, are
there any good reasons to develop a replacement system?
Keith.
sufficiently robust for ActiveState and others to run on public servers.
When they were first released, it took a little while to get used to them;
but they quickly became indispensable. Now that they are open-source, are
there any good reasons to develop a replacement system?
Keith.
Ali M
Jan 16, 2018, 11:23:18 PM1/16/18
to
On Tuesday, January 16, 2018 at 10:50:36 PM UTC-5, Keith Nash wrote:
> How much development do Teacup/Teapot need? They are mature code,
> sufficiently robust for ActiveState and others to run on public servers.
>
> When they were first released, it took a little while to get used to them;
> but they quickly became indispensable. Now that they are open-source, are
> there any good reasons to develop a replacement system?
>
> Keith.
>
I am not sure how the tcl community works, but most communities especially small ones (and i consider tcl to be a small one) are led by few, usually the creator
> How much development do Teacup/Teapot need? They are mature code,
> sufficiently robust for ActiveState and others to run on public servers.
>
> When they were first released, it took a little while to get used to them;
> but they quickly became indispensable. Now that they are open-source, are
> there any good reasons to develop a replacement system?
>
> Keith.
>
some even argue that tcl unpopularity is because it is missing a bdfl (benevolent dictator for life)
Anyway, i think this needs a push from the core team
the core team need to push for those tools or a replacement
Active state seems to be happy sponsoring a host for the service
jsunth...@gmail.com
Jan 17, 2018, 9:34:22 AM1/17/18
to
So, lets just round it down, way down, and say 20 years, two decades. Maybe it's been closer to a quarter of century, but we'll call it 20 years for our purposes here.
It turns out that there is actually a wiki category "Repository", dedicated to the fits and starts and quiet deaths of a multitude of Tcl CPANs.
Tcl is not left with nothing: We have tcllib and tklib, a blessed and curated collection of Tcl packages- I mean, they aren't the size of CPAN or NPM, but they exist. We have the wiki, which itself is a copy-and-paste-based source repository. A big Huzzah to all who posted their code in the wiki rather than providing a link to their personal high performance gopher:// server with which to offer their code to the.. uh.. masses, lets just say. Linux distros have their package managers and tcl packages available in them, windows has ActiveTcl (or you can use it on mac and linux), and I know absolutely nothing further about the appleverse, but I'm sure there a multitude of ways there too. And despite the Tcl communities' ethic of insular obscurity, the kids are posting Tcl code on github, 'cuz that's just what you do, as far as they know.
I just read yesterday, before reading the above response that also mentions it, that ActiveState has open sourced the teapot/cup system, but in fairness, it was rather recently, and word had not spread to my ear. Yesterday I also discovered
http://teaparty.rkeene.org/fossil/home
http://teapot.rkeene.org/index.html
Now, I'm no big city lawyer, but I do believe that's an opensource CPAN just kinda, you know, hangin' out. Chillin'.
I want to apologize for propagating this notion that the teapot/cup thing was dead. I asked about it in the irc channel, and was directed away from it from somebody or another, and I accepted that as cannon, having just returned from the JavaScript mines. I don't use ActiveTcl, so I just didn't know.
At the bottom of https://wiki.tcl.tk/16925 Tcl's own DFK, in response to his own comp.lang.tcl rant circa 2010, wrote a pure tcl package grabbing client, directed at the fictional "http://repository.tcl.tk/packages" URL. That would be cool, huh?
Of course, there is not even an easy answer in Tcl to the question, "What's a Tcl package?" ActiveState drew the line at "these 4 things" for the teapot/cup, apparently, but acknowledges that it isn't really a comprehensive picture of the landscape.
======
So my own spin:
Tcl was created to be a layer on top of binary systems, so lots of the goodies in Tcl aren't pure tcl. NPM sometimes needs to compile stuff, and does so surprisingly reliably.
Tcl's own Hypnotoad has a system called ODIE, which is a sort of mother-of-all-build systems, written in his homegrown systems called Tao and Practcl, and uses Fossil. If something about that is untrue, it's only due to confusion on my part. That's the best explanation I can give. What I can say is that on a good day, one can use ODIE to build a binary tclkit of 8.6.8, and a "toadkit" containing a bunch of dank stuff like tDOM. If 8.7a2/9 was building on Linux, it could use that too. And although I'm not sure if it's all considered ODIE or not, there are a bunch of Fossil repos in it's general vicinity with more Tcl extensions to be had.
An easy-reader version of Tao is found in tcllib 1.18, called "Tool", and a version of Practcl is there too. I posted on the wiki recently how to build Fossil with Tcl integration on Debian; it's a piece of cake.
THE TCLERS DOBEE
Imagine a world where there was no gatekeeper for the Tcl's CPAN, just like the The Tclers Wiki, where crying out into the desert night for the attention of some Tcl Illuminati was not required to include a package. Imagine this CPAN could build binary extensions, and offer pre-compiled packages and bundle ToadTclStarKitPack-thingies.
- Individuals could roll what they want into The Dobee.
- The Dobee could be passed one to another.
- The Dobee requires no special client. If it can HTTP, it can take a drag, I mean, pull from The Dobee.
The Doobe would complement the The Wiki, and share the same ethos. Aloha!
Sure, it's more than likely to be an also-ran NPM for Tcl. But it hasn't been tried before. And we have almost all the parts we need working already.
Keith Nash
Jan 26, 2018, 2:34:49 PM1/26/18
to
IMHO any repository needs a gatekeeper (or group of gatekeepers) - otherwise
there is nothing to stop people uploading malicious code, especially
malicious binaries.
I don't use Perl, but I never hear anything good about the plethora of
packages in CPAN.
The problem with any Tcl repository seems to be that it works well when it
is launched, but it gradually becomes out of date if the maintainer is busy
with other things.
Perhaps a repository could include a cron job that checks for new releases
of each package that it provides, fetches them, builds them, tests them, and
installs them. Binary packages would ideally include a cross-platform build
script that need not be completely general, but would be suitable for
generating a repository package. This would automate some of the work of
maintaining the repository.
The BSD "ports" systems might include code that we could adapt for our use
(without using "ports" itself as yet another repository system!).
Keith.
there is nothing to stop people uploading malicious code, especially
malicious binaries.
I don't use Perl, but I never hear anything good about the plethora of
packages in CPAN.
The problem with any Tcl repository seems to be that it works well when it
is launched, but it gradually becomes out of date if the maintainer is busy
with other things.
Perhaps a repository could include a cron job that checks for new releases
of each package that it provides, fetches them, builds them, tests them, and
installs them. Binary packages would ideally include a cross-platform build
script that need not be completely general, but would be suitable for
generating a repository package. This would automate some of the work of
maintaining the repository.
The BSD "ports" systems might include code that we could adapt for our use
(without using "ports" itself as yet another repository system!).
Keith.
Robert
Jan 27, 2018, 4:30:10 PM1/27/18
to
Like anything you have to weed out the chaff and CPAN is no different. CPAN still works and it works well. Tcl would benefit for a similar ecosystem. MacPorts is all Tcl from what I can tell. You could probably leverage a lot of that too.
So you would need:
* somewhere to host
* make it easy to propagate repos
* have some kind of checksum system
* be able to query, install, upgrade, and remove packages
--
Bob
So you would need:
* somewhere to host
* make it easy to propagate repos
* have some kind of checksum system
* be able to query, install, upgrade, and remove packages
--
Bob
0 new messages