Was a consensus reached re: HTTP/HTTPS?
Locked
 
|
In scanning through the list archives I see that there was some
discussion back 2012 about picking a HTTP client. It looks like Zinc and WebClient were the front-runners but I don't see anything more recent indicating a decision so just figured I'd ask where things stood on this front... _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
|
Hi Phil: I evaluated some time ago the effort involved in port Zinc and WebClient. Zinc is very tied to Pharo and would need a lot of work to make it work in Cuis (and, of course, to maintain it in sync with the Pharo version). Also I forked WebClient of a preliminar port from David Graham and did some updates and so and the client part worked (the last time that I tried) but not the server part. The port of the server part also involves a lot of work and, as I was not able to sell a commercial solution that make me earn some money to justify the work, I not tried anymore. Cheers. 2015-02-24 22:02 GMT-03:00 Phil (list) <[hidden email]>: In scanning through the list archives I see that there was some Saludos / Regards, www.arduinosoftware.comGermán Arduino _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
|
On Wed, 2015-03-04 at 17:24 -0300, Germán Arduino wrote:
> Hi Phil: > Hi Germán, > > I evaluated some time ago the effort involved in port Zinc and > WebClient. > > > Zinc is very tied to Pharo and would need a lot of work to make it > work in Cuis (and, of course, to maintain it in sync with the Pharo > version). > > > Also I forked WebClient of a preliminar port from David Graham and did > some updates and so and the client part worked (the last time that I > tried) but not the server part. The port of the server part also > involves a lot of work and, as I was not able to sell a commercial > solution that make me earn some money to justify the work, I not tried > anymore. > haven't merged back into Davids repo but both appear to have been updated relatively recently (to each other) so I'm curious: what were the main changes you needed to make in your fork? Given the lack of anyone willing to take on porting Zinc, I'm thinking I'll just use one of the ports you guys have already done but am not sure which one I should be looking at... > Cheers. > Thanks, Phil > > > 2015-02-24 22:02 GMT-03:00 Phil (list) <[hidden email]>: > In scanning through the list archives I see that there was > some > discussion back 2012 about picking a HTTP client. It looks > like Zinc > and WebClient were the front-runners but I don't see anything > more > recent indicating a decision so just figured I'd ask where > things stood > on this front... > > > _______________________________________________ > Cuis mailing list > [hidden email] > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org > > > > > -- > Saludos / Regards, > Germán Arduino > > www.arduinosoftware.com > > > _______________________________________________ > Cuis mailing list > [hidden email] > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
|
In reply to this post by Phil B
Hi Phil,
I just pulled down the Cuis4.2-2203 along with Cog 3266, and the WebClient package seems to work fine with both the client and server examples on my Mac (Sorry, I still haven’t had a chance to get back to work on the tests). You will want to install the Cuis Core-Packages, then Cuis-DigitalSignatures, Cuis-UUID, and finally the Cuis-WebClient packages from my repo. garduino, Sorry, I took a look at your pull request and it’s adds 8K LOC to a 5K LOC project. I’d like to keep this as close to the Cuis base as possible and avoid compatibility libraries. Thanks, David _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
|
On Sun, 2015-03-08 at 10:18 -0500, David Graham wrote:
> Hi Phil, > > I just pulled down the Cuis4.2-2203 along with Cog 3266, and the WebClient package seems to work fine with both the client and server examples on my Mac (Sorry, I still haven’t had a chance to get back to work on the tests). You will want to install the Cuis Core-Packages, then Cuis-DigitalSignatures, Cuis-UUID, and finally the Cuis-WebClient packages from my repo. > Thanks for commenting David. That's what I was trying to understand... WebClient from your repo seems to be working as expected for me. Since HTTP client/server code is a pretty key bit of functionality for many types of apps and there doesn't seem to be the time/interest in working on Zinc right now, should we look at making this a core package? (of course, we could always switch in the future if that makes sense down the road) > garduino, Sorry, I took a look at your pull request and it’s adds 8K LOC to a 5K LOC project. I’d like to keep this as close to the Cuis base as possible and avoid compatibility libraries. > We're a small enough community that divergent forks are not a great idea if we can avoid them. I've already retired my FFI port and am doing the same with my XML-Parser port as core packages now exist for both. It would be great if we could try to keep to a common code base where possible... > Thanks, > David > _______________________________________________ > Cuis mailing list > [hidden email] > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
Re: Was a consensus reached re: HTTP/HTTPS?
|
|
+1 for having a web package in
https://github.com/Cuis-Smalltalk/Cuis-Smalltalk-Dev/tree/master/Packages Has a consensus ben reached? On 3/8/15, Phil (list) <[hidden email]> wrote: > On Sun, 2015-03-08 at 10:18 -0500, David Graham wrote: >> Hi Phil, >> >> I just pulled down the Cuis4.2-2203 along with Cog 3266, and the WebClient >> package seems to work fine with both the client and server examples on my >> Mac (Sorry, I still haven’t had a chance to get back to work on the >> tests). You will want to install the Cuis Core-Packages, then >> Cuis-DigitalSignatures, Cuis-UUID, and finally the Cuis-WebClient packages >> from my repo. >> > > Thanks for commenting David. That's what I was trying to understand... > WebClient from your repo seems to be working as expected for me. Since > HTTP client/server code is a pretty key bit of functionality for many > types of apps and there doesn't seem to be the time/interest in working > on Zinc right now, should we look at making this a core package? (of > course, we could always switch in the future if that makes sense down > the road) > >> garduino, Sorry, I took a look at your pull request and it’s adds 8K LOC >> to a 5K LOC project. I’d like to keep this as close to the Cuis base as >> possible and avoid compatibility libraries. >> > > We're a small enough community that divergent forks are not a great idea > if we can avoid them. I've already retired my FFI port and am doing the > same with my XML-Parser port as core packages now exist for both. It > would be great if we could try to keep to a common code base where > possible... > >> Thanks, >> David >> _______________________________________________ >> Cuis mailing list >> [hidden email] >> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org > > > > _______________________________________________ > Cuis mailing list > [hidden email] > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org > _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
|
In reply to this post by dsg
2015-03-08 12:18 GMT-03:00 David Graham <[hidden email]>:
Hi David, not problem at all, but I'm curious how it works without the compatibility layer? You modified the WebClient itself? And about the server, it works? Last time that I tried didn't worked. Thanks! _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
|
In reply to this post by Hannes Hirzel
On Mon, 2015-03-09 at 11:44 +0000, H. Hirzel wrote:
> +1 for having a web package in > https://github.com/Cuis-Smalltalk/Cuis-Smalltalk-Dev/tree/master/Packages > > Has a consensus ben reached? > It seems a defacto consensus has been reached: the original conversation occurred about 3 years ago and the only option that appears to exist as of today is WebClient. _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
Re: Was a consensus reached re: HTTP/HTTPS?
|
|
Final word from Juan?
On 3/9/15, Phil (list) <[hidden email]> wrote: > On Mon, 2015-03-09 at 11:44 +0000, H. Hirzel wrote: >> +1 for having a web package in >> https://github.com/Cuis-Smalltalk/Cuis-Smalltalk-Dev/tree/master/Packages >> >> Has a consensus ben reached? >> > > It seems a defacto consensus has been reached: the original conversation > occurred about 3 years ago and the only option that appears to exist as > of today is WebClient. > > > > _______________________________________________ > Cuis mailing list > [hidden email] > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org > _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
|
I’d be honored to see the WebClient port included in the core Packages and I’m fine with any decision Juan makes. I really like the Cuis philosophy, so most of the porting involved rewriting code to conform to what is available in Cuis.
Just a couple notes: 1. The Cuis-DigitalSignatures and Cuis-UUID packages in my repo should be imported as well. These invoke VM primitives and should remain independent. 2. I've used the the WebClient code on my own, without trouble, but the unit tests are very broken and I haven’t had enough spare time to fix these. Also, the Cuis-SecureSocket package in my repo is a port of the SqueakSSL code. This code works, but needs a lot of attention. I had a trivial Cuis update app that monitored and could install CoreUpdates directly from github, but github switched to using TLS SNI and the SNI update hasn’t made its way into the Squeak VM plugin yet: http://bugs.squeak.org/view.php?id=7824 > On Mar 9, 2015, at 1:19 PM, H. Hirzel <[hidden email]> wrote: > > Final word from Juan? > > On 3/9/15, Phil (list) <[hidden email]> wrote: >> On Mon, 2015-03-09 at 11:44 +0000, H. Hirzel wrote: >>> +1 for having a web package in >>> https://github.com/Cuis-Smalltalk/Cuis-Smalltalk-Dev/tree/master/Packages >>> >>> Has a consensus ben reached? >>> >> >> It seems a defacto consensus has been reached: the original conversation >> occurred about 3 years ago and the only option that appears to exist as >> of today is WebClient. >> >> >> >> _______________________________________________ >> Cuis mailing list >> [hidden email] >> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org >> > > _______________________________________________ > Cuis mailing list > [hidden email] > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
|
On Mon, 2015-03-09 at 14:04 -0500, David Graham wrote:
> I’d be honored to see the WebClient port included in the core Packages and I’m fine with any decision Juan makes. I really like the Cuis philosophy, so most of the porting involved rewriting code to conform to what is available in Cuis. > > Just a couple notes: > 1. The Cuis-DigitalSignatures and Cuis-UUID packages in my repo should be imported as well. These invoke VM primitives and should remain independent. > 2. I've used the the WebClient code on my own, without trouble, but the unit tests are very broken and I haven’t had enough spare time to fix these. > > Also, the Cuis-SecureSocket package in my repo is a port of the SqueakSSL code. This code works, but needs a lot of attention. I had a trivial Cuis update app that monitored and could install CoreUpdates directly from github, but github switched to using TLS SNI and the SNI update hasn’t made its way into the Squeak VM plugin yet: > http://bugs.squeak.org/view.php?id=7824 > One other thing that would probably need to be done is to split out the tests into a separate package(s). If it is decided to go with WebClient, I can help with fixing up the tests. _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
Re: Was a consensus reached re: HTTP/HTTPS?
|
|
Splitting up the tests into a separate package is a good thing in any
case. Please just go ahead with it. Then fixing the tests is an issue of quality assurance for the WebClient package. On 3/9/15, Phil (list) <[hidden email]> wrote: > On Mon, 2015-03-09 at 14:04 -0500, David Graham wrote: >> I’d be honored to see the WebClient port included in the core Packages and >> I’m fine with any decision Juan makes. I really like the Cuis philosophy, >> so most of the porting involved rewriting code to conform to what is >> available in Cuis. >> >> Just a couple notes: >> 1. The Cuis-DigitalSignatures and Cuis-UUID packages in my repo should be >> imported as well. These invoke VM primitives and should remain >> independent. >> 2. I've used the the WebClient code on my own, without trouble, but the >> unit tests are very broken and I haven’t had enough spare time to fix >> these. >> >> Also, the Cuis-SecureSocket package in my repo is a port of the SqueakSSL >> code. This code works, but needs a lot of attention. I had a trivial >> Cuis update app that monitored and could install CoreUpdates directly from >> github, but github switched to using TLS SNI and the SNI update hasn’t >> made its way into the Squeak VM plugin yet: >> http://bugs.squeak.org/view.php?id=7824 >> > > One other thing that would probably need to be done is to split out the > tests into a separate package(s). If it is decided to go with > WebClient, I can help with fixing up the tests. > > > _______________________________________________ > Cuis mailing list > [hidden email] > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org > _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
Re: Was a consensus reached re: HTTP/HTTPS?
|
|
In reply to this post by dsg
Hi Folks,
WRT optional packages in the main Cuis repo, I don't think it is my decision. I think that what is needed is consensus between people who know and care about the options, pros and cons, etc. All I ask is that no code in the main Cuis repo is abandoned. I mean, the moment we know that something is badly broken, and nobody will work on fixing it in some reasonable time, that's the time to remove that code. In particular, wrt HTTP/HTTPS, I don't have much knowledge about the options, and I'm not using any of them. So, I prefer you guys to make the decision. So, the proposal is to integrate into the mail Cuis repo, the packages at: - https://github.com/davidgraham/Cuis-WebClient - https://github.com/davidgraham/Cuis-UUID - https://github.com/davidgraham/Cuis-DigitalSignatures Please tell: - Do you think this is the best option? - Or you think it is not? (why? which would be better?) - Or (like me) you don't really know, and you prefer those who know better decide? Let's give naysayers a chance to speak, and if we agree, I'll merge them. Cheers, Juan Vuletich On 3/9/2015 4:04 PM, David Graham wrote: > I’d be honored to see the WebClient port included in the core Packages and I’m fine with any decision Juan makes. I really like the Cuis philosophy, so most of the porting involved rewriting code to conform to what is available in Cuis. > > Just a couple notes: > 1. The Cuis-DigitalSignatures and Cuis-UUID packages in my repo should be imported as well. These invoke VM primitives and should remain independent. > 2. I've used the the WebClient code on my own, without trouble, but the unit tests are very broken and I haven’t had enough spare time to fix these. > > Also, the Cuis-SecureSocket package in my repo is a port of the SqueakSSL code. This code works, but needs a lot of attention. I had a trivial Cuis update app that monitored and could install CoreUpdates directly from github, but github switched to using TLS SNI and the SNI update hasn’t made its way into the Squeak VM plugin yet: > http://bugs.squeak.org/view.php?id=7824 > > >> On Mar 9, 2015, at 1:19 PM, H. Hirzel<[hidden email]> wrote: >> >> Final word from Juan? >> >> On 3/9/15, Phil (list)<[hidden email]> wrote: >>> On Mon, 2015-03-09 at 11:44 +0000, H. Hirzel wrote: >>>> +1 for having a web package in >>>> https://github.com/Cuis-Smalltalk/Cuis-Smalltalk-Dev/tree/master/Packages >>>> >>>> Has a consensus ben reached? >>>> >>> It seems a defacto consensus has been reached: the original conversation >>> occurred about 3 years ago and the only option that appears to exist as >>> of today is WebClient. >>> >>> >>> >>> _______________________________________________ >>> Cuis mailing list >>> [hidden email] >>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org >>> >> _______________________________________________ >> Cuis mailing list >> [hidden email] >> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org > > _______________________________________________ > Cuis mailing list > [hidden email] > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
On Tue, 10 Mar 2015 10:00:09 -0300
Juan Vuletich <[hidden email]> wrote: > In particular, wrt HTTP/HTTPS, I don't have much knowledge about the > options, and I'm not using any of them. So, I prefer you guys to make > the decision. ... > - Or (like me) you don't really know, and you prefer those who know > better decide? Not a user. No informed opinion. -KenD _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
-KenD
|
Re: Was a consensus reached re: HTTP/HTTPS?
|
|
I use WebClient on Squeak. Not used so far with Cuis. Please go ahead
with the proposal. --Hannes On 3/10/15, Ken.Dickey <[hidden email]> wrote: > On Tue, 10 Mar 2015 10:00:09 -0300 > Juan Vuletich <[hidden email]> wrote: > >> In particular, wrt HTTP/HTTPS, I don't have much knowledge about the >> options, and I'm not using any of them. So, I prefer you guys to make >> the decision. > ... > >> - Or (like me) you don't really know, and you prefer those who know >> better decide? > > Not a user. No informed opinion. > > -KenD > > _______________________________________________ > Cuis mailing list > [hidden email] > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org > _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
|
Mine was a different approach, I tried to make the minimun possible changes to WC itself and rely on a compatibility layer, to make more easy the follow up of the maintenance of the official release. But if the current packages from David are working I'm not against of the propossed approach and I will delete my fork.2015-03-10 11:28 GMT-03:00 H. Hirzel <[hidden email]>: I use WebClient on Squeak. Not used so far with Cuis. Please go ahead -- Saludos / Regards, www.arduinosoftware.comGermán Arduino _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
Re: Was a consensus reached re: HTTP/HTTPS?
|
|
Yes, Germán's approach has its merit as well. It was to avoid forking
WebClient as much as possible and thus relying on a compatibility layer. In any case WebClient for Cuis is a fork and needs to be maintained. But this applies for all of the packages in the common repository. I remember that the late Andreas Raab wrote he did about 3 to 4 attempts until he came up with WebClient. Zinc then was a fork of it at that time. --Hannes On 3/10/15, Germán Arduino <[hidden email]> wrote: > Mine was a different approach, I tried to make the minimun possible changes > to WC itself and rely on a compatibility layer, to make more easy the > follow up of the maintenance of the official release. > > But if the current packages from David are working I'm not against of the > propossed approach and I will delete my fork. > > > 2015-03-10 11:28 GMT-03:00 H. Hirzel <[hidden email]>: > >> I use WebClient on Squeak. Not used so far with Cuis. Please go ahead >> with the proposal. >> >> --Hannes >> >> On 3/10/15, Ken.Dickey <[hidden email]> wrote: >> > On Tue, 10 Mar 2015 10:00:09 -0300 >> > Juan Vuletich <[hidden email]> wrote: >> > >> >> In particular, wrt HTTP/HTTPS, I don't have much knowledge about the >> >> options, and I'm not using any of them. So, I prefer you guys to make >> >> the decision. >> > ... >> > >> >> - Or (like me) you don't really know, and you prefer those who know >> >> better decide? >> > >> > Not a user. No informed opinion. >> > >> > -KenD >> > >> > _______________________________________________ >> > Cuis mailing list >> > [hidden email] >> > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org >> > >> >> _______________________________________________ >> Cuis mailing list >> [hidden email] >> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org >> > > > > -- > Saludos / Regards, > Germán Arduino > www.arduinosoftware.com > _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
|
|
In reply to this post by Juan Vuletich-4
On Tue, 2015-03-10 at 10:00 -0300, Juan Vuletich wrote:
> Hi Folks, > > WRT optional packages in the main Cuis repo, I don't think it is my > decision. I think that what is needed is consensus between people who > know and care about the options, pros and cons, etc. All I ask is that > no code in the main Cuis repo is abandoned. I mean, the moment we know > that something is badly broken, and nobody will work on fixing it in > some reasonable time, that's the time to remove that code. > > In particular, wrt HTTP/HTTPS, I don't have much knowledge about the > options, and I'm not using any of them. So, I prefer you guys to make > the decision. > > So, the proposal is to integrate into the mail Cuis repo, the packages at: > - https://github.com/davidgraham/Cuis-WebClient > - https://github.com/davidgraham/Cuis-UUID > - https://github.com/davidgraham/Cuis-DigitalSignatures > > Please tell: > - Do you think this is the best option? > - Or you think it is not? (why? which would be better?) > - Or (like me) you don't really know, and you prefer those who know > better decide? > This seems like the only viable option at this time to me... I'm in the process of migrating my HTTP code over to it. As Germán mentioned, performing and maintaining a port of Pharo code (i.e. Zinc) would be challenging and not something I'd want to take on. > Let's give naysayers a chance to speak, and if we agree, I'll merge them. > I'm all for hearing any arguments against if anyone thinks there's a better solution... > Cheers, > Juan Vuletich > > On 3/9/2015 4:04 PM, David Graham wrote: > > I’d be honored to see the WebClient port included in the core Packages and I’m fine with any decision Juan makes. I really like the Cuis philosophy, so most of the porting involved rewriting code to conform to what is available in Cuis. > > > > Just a couple notes: > > 1. The Cuis-DigitalSignatures and Cuis-UUID packages in my repo should be imported as well. These invoke VM primitives and should remain independent. > > 2. I've used the the WebClient code on my own, without trouble, but the unit tests are very broken and I haven’t had enough spare time to fix these. > > > > Also, the Cuis-SecureSocket package in my repo is a port of the SqueakSSL code. This code works, but needs a lot of attention. I had a trivial Cuis update app that monitored and could install CoreUpdates directly from github, but github switched to using TLS SNI and the SNI update hasn’t made its way into the Squeak VM plugin yet: > > http://bugs.squeak.org/view.php?id=7824 > > > > > >> On Mar 9, 2015, at 1:19 PM, H. Hirzel<[hidden email]> wrote: > >> > >> Final word from Juan? > >> > >> On 3/9/15, Phil (list)<[hidden email]> wrote: > >>> On Mon, 2015-03-09 at 11:44 +0000, H. Hirzel wrote: > >>>> +1 for having a web package in > >>>> https://github.com/Cuis-Smalltalk/Cuis-Smalltalk-Dev/tree/master/Packages > >>>> > >>>> Has a consensus ben reached? > >>>> > >>> It seems a defacto consensus has been reached: the original conversation > >>> occurred about 3 years ago and the only option that appears to exist as > >>> of today is WebClient. > >>> > >>> > >>> > >>> _______________________________________________ > >>> Cuis mailing list > >>> [hidden email] > >>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org > >>> > >> _______________________________________________ > >> Cuis mailing list > >> [hidden email] > >> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org > > > > _______________________________________________ > > Cuis mailing list > > [hidden email] > > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org > > > _______________________________________________ > Cuis mailing list > [hidden email] > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org _______________________________________________ Cuis mailing list [hidden email] http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org |
«
Return to Cuis Smalltalk
|
1 view|%1 views
| Free forum by Nabble | Edit this page |