July 2021 - Pharo-users - lists.pharo.org

Re: ZnServer on a cheap server: your experiences

by Yanni Chiu

Yes, very stable. I've had a DigitalOcean droplet running for 6 years that was holding various false starts. To my surprise, one (never "shipped") website was still running (Pharo-3.?, mongodb, Bootstrap css). I'd been poking the site every once in a while, and found that the Bootstrap UI steadily degraded, because I'd pointed the site's .css at some non-fixed version from some CDN server (so that .css must have changed over the years). The uptime on the machine was over 1000 days, and there have been many emails over the years about network, hardware, etc. changes. After years of unattended reboots, some relic was still up. Anyhow, that instance is now deleted, probably soon to be joined by a more recent setup where I started with the DO 1-click docker droplet. This setup was useful to learn about docker, and I might go back to it in the future. Given Estaban's comments in this thread, I decided to do a bare nginx + Pharo image deployment (which is similar to the now deleted 6 year instance described above, which had Apache2 instead of nginx). I will have the mongo server on a separate machine, but have not decided whether to use the DO offering of a managed mongo server. It's still a work in progress, and other's experiences are of interest. Thanks for reading. Yanni Chiu On Thu, Jul 22, 2021 at 8:33 AM Esteban Maringolo <emaringolo(a)gmail.com> wrote: > > Yeap, not much more to add than what Sven said. What's important to > mention is that they are really stable. > I've been running Pharo servers in DO droplets of all sizes, without > issues (some running for months), I only had to upscale one droplet to > more memory, and it was because of a leak I introduced with PGSQL > connections, otherwise, they're pretty lightweight for today's > standards and a normal workload. > > I have deployments with nginx and Pharo images as upstreams, and I > have one with Docker swarm and Traefik doing the load balancing among > different Pharo workers and acting as the HTTPS endpoint. I'm removing > this option though and moving back to nginx only. > > What I never tried was to host a docker container running Pharo, in > some Docker hosting. It might be the best option as a quick start, but > after doing the math, it's always more expensive than a Droplet. > > Best regards, > > Esteban A. Maringolo > > On Thu, Jul 22, 2021 at 3:21 AM Sven Van Caekenberghe <sven(a)stfx.eu> wrote: > > > > Hi Vince, > > > > That is certainly possible and works well. I would recommend an instance with 1GB RAM, that leaves you some headroom. > > > > Deploying web applications is of course a broad subject, much of the required knowledge is not Pharo specific, but needed anyway. > > > > The last chapter in the Pharo Enterprise book (https://books.pharo.org/enterprise-pharo/) is a good starting point (Deployment). But there are other and different approaches. > > > > A plain HTTP demo instance running on a DO instance can be found here: http://zn.stfx.eu/welcome > > > > For production use you should front with NGINX or something similar to add HTTPS. > > > > May people on this mailing list deploy Pharo server applications, we have tens of them in day to day production doing real work. > > > > Good luck on your journey, you know where to ask questions. > > > > Regards, > > > > Sven > > > > > On 22 Jul 2021, at 07:56, vinref(a)gmail.com wrote: > > > > > > Anyone here run a web app using plain ZnServer (or subclasses) on a cheap VPS (i.e., $10/month DO droplet or equivalent). What are your experiences?, suggestions. > > > > > > I am planning a web app with just plain ZnServer, SQLite3, ATS or equivalent. > > > > > > Thanks, Vince > > >

July 26, 2021

Re: Error in Pharo 9's Spec UI Framework Demo

by tau

Thank a lot for directing me toward SpFilteringListPresenter. I think I found a bug in SpFilteringSelectableListPresenter. While using an SpFilteringSelectableListPresenter, if you press ctrl-a, every item in the list will get selected. Then, if you deselect a few items and press ctrl-a again, this time nothing happens. However, if you scroll the list, all the items are suddenly selected again. This happens because the block that get invoked when ctrl-a is pressed doesn't refresh the presenter. When I changed that block from [ self activateAll ] to [ self activateAll. listPresenter refresh ], the bug disappeared. ________________________________ De : Esteban Lorenzano <estebanlm(a)netc.eu> EnvoyÃ© : 25 juillet 2021 02:28:51 Ã : Any question about pharo is welcome Cc : pharo-users(a)lists.pharo.org Objet : [Pharo-users] Re: Error in Pharo 9's Spec UI Framework Demo Hi Tau, The filter that was bundled with the list components is deprecated and should not be used. Instead, we provide another component with that functionality: SpFilteringListPresenter. Take a loot there and its examples. Esteban On Jul 25 2021, at 5:52 am, tau <tau(a)cedalion.info> wrote: In the Spec2 examples included with Pharo9, in the example for SpListPresenter, when I try to input text in the list filter, I get an error: the debug window says: "Instance of OrderedCollection did not understand #hasElementAt:". The example works correctly in Pharo8. Is there a workaround to make the example work?

July 26, 2021

Re: R: Re: The Greatest Contributors to Smalltalk since 1980

by horrido.hobbies＠gmail.com

With very few exceptions, no important programming language ever âdies.â Even APL and Forth are still in use today, if only in a small way (including their derivatives like J and Factor). But from the perspective of the programming community at large, if a programming language no longer has a large user base, itâs effectively dead. This may not be fair but there you have it. We may argue that it doesnât matter. If Smalltalk serves a small community well, thatâs the important thing. I take a broader perspective. We live in a technological world where software development productivity is a serious bottleneck. Smalltalk has the potential to greatly alleviate this bottleneck. Cutting development time in half (or better) matters a lot. This, I believe, is what we must fight for. We may fail. Indeed, we *probably* *will fail*. But if we donât try, we definitely will fail. Cheers,\ Richard Lorenzo wrote: > You are absolutely right! > > Lorenzo > > \-----Messaggio originale----- > Da: Esteban Maringolo \[mailto:emaringolo@gmail.com\] > Inviato: domenica 25 luglio 2021 17:54 > A: Any question about pharo is welcome [pharo-users@lists.pharo.org](mailto:pharo-users@lists.pharo.org) > Oggetto: \[Pharo-users\] Re: The Greatest Contributors to Smalltalk since 1980 > > On Sun, Jul 25, 2021 at 11:31 AM Tim Mackinnon [tim@testit.works](mailto:tim@testit.works) wrote: > > > Isnât this the wrong question to ask? Iâm assuming this is to do with Smalltalkâs 50th anniversary, and of course we are grateful to those early pioneers who did lots of work in the field 20-30 years ago but to me thatâs the old history and while itâs interesting to call out, it doesnât shed life on the day to day energy we have today - whst keeps Smalltalk alive and current. > > Hi share the view, it's the wrong question, and it pursues that "hero worshipping" culture that is already dead (or at least outdated) since a long time ago. > > Additionally, I don't share the "keeping Smalltalk alive" expression, as if dying was its inevitable outcome. I haven't heard "keeping LISP alive" (and I don't call LISP as dead either). It would be self-deceiving to call ourselves mainstream, but that doesn't mean we're doomed somehow. > > So in 50 years we should celebrate the half-century, remember the history, look at what we did "wrong", and focus on looking forward, because "the best way to predict the future is to invent it" ;-) > > Regards! > > Esteban A. Maringolo

July 26, 2021

R: Re: The Greatest Contributors to Smalltalk since 1980

by Lorenzo

You are absolutely right! Lorenzo -----Messaggio originale----- Da: Esteban Maringolo [mailto:emaringolo@gmail.com] Inviato: domenica 25 luglio 2021 17:54 A: Any question about pharo is welcome <pharo-users(a)lists.pharo.org> Oggetto: [Pharo-users] Re: The Greatest Contributors to Smalltalk since 1980 On Sun, Jul 25, 2021 at 11:31 AM Tim Mackinnon <tim(a)testit.works> wrote: > Isnât this the wrong question to ask? Iâm assuming this is to do with Smalltalkâs 50th anniversary, and of course we are grateful to those early pioneers who did lots of work in the field 20-30 years ago but to me thatâs the old history and while itâs interesting to call out, it doesnât shed life on the day to day energy we have today - whst keeps Smalltalk alive and current. Hi share the view, it's the wrong question, and it pursues that "hero worshipping" culture that is already dead (or at least outdated) since a long time ago. Additionally, I don't share the "keeping Smalltalk alive" expression, as if dying was its inevitable outcome. I haven't heard "keeping LISP alive" (and I don't call LISP as dead either). It would be self-deceiving to call ourselves mainstream, but that doesn't mean we're doomed somehow. So in 50 years we should celebrate the half-century, remember the history, look at what we did "wrong", and focus on looking forward, because "the best way to predict the future is to invent it" ;-) Regards! Esteban A. Maringolo

July 26, 2021

Re: Zinc exception logging

by Esteban Maringolo

Hi Sven, That's what I wanted to know, I made a quick change to it and pushed it to the staging server (where I'm having the error), but I forgot to simply search for the string, and now I confirm this is returned by Seaside's response generator in the exception handler. I don't know where the 4 comes from, and it only happens in production, but replacing it with another exception handler gives me a better understanding of what might be going on. It seems there was a change in some dependency (seems like Glorp) that's affecting my staging server. Thanks! Esteban A. Maringolo On Mon, Jul 26, 2021 at 3:16 AM Sven Van Caekenberghe <sven(a)stfx.eu> wrote: > > Hi Esteban, > > > On 26 Jul 2021, at 02:47, Esteban Maringolo <emaringolo(a)gmail.com> wrote: > > > > Is there a way to have a "stack dump" response in Zinc? > > > > #debugMode is good for development, but I'm having an "Internal Error: > > 4", that I don't know how to trace. > > > > * Trying 167.71.182.110... > > * TCP_NODELAY set > > * Connected to fore.base.golf (167.71.182.110) port 8090 (#0) > >> GET /web HTTP/1.1 > >> Host: x.y.z:8090 > >> User-Agent: curl/7.58.0 > >> Accept: */* > >> > > < HTTP/1.1 500 Internal Server Error > > < Content-Length: 512 > > < Content-Type: text/plain;charset=utf-8 > > < Server: Zinc HTTP Components 1.0 (Pharo/8.0) > > < Date: Mon, 26 Jul 2021 00:45:39 GMT > > < > > * Connection #0 to host x.y.z left intact > > Internal Error: 4 > > > > Regards! > > > > Esteban A. Maringolo > > Searching the source code of my image, I only find 'Internal Error' in WAExceptionHandler/WAResponseGenerator, not in Zn code. Furthermore, printing 4 there would be weird (as it is the exception message). > > Is this a plain Zn response or a Seaside response ? > > I know that Seaside has more logging options. > > Zn does not have an exception logging mechanism that dumps a stack trace, only the debugMode for interactive development. I see you used curl, can't you fire the request against your dev image then ? > > Sven

July 26, 2021

Re: The Greatest Contributors to Smalltalk since 1980

by jtuchel＠objektfabrik.de

I guess I agree on the question being wrong. There are quiet a few people I think did incredibly important things, not only to Smalltalk. What about the people behind the Refactoring Browser (John Brant? and I guess a few less prominent people who did some of the groundwork), for example? Seaside has been mentioned already, but there are other milestones, like TOPLink or Glorp (Alan Knight and his ex-colleagues). Then there are people who did a lot of ground work that tends to be unseen. Niall Ross has been working in SUnit, RB, Rewriet Tool and Glorp literally for decades without getting any laurels because he "only" carried these milestones forward and kept them working. And there are quite a few people who did amazing things that we tend to forget about. The success of Smalltalk (and even if we are not mainstream, our language has survived for 50 years and will not go anywhere soon, some hyped technologies won't even make 10, not to think of 20 years) is based on the work of so many people. What about Vanessa Freudenberg, The whole Squeak and EToys community, OpenQwaq and scratch? OLPC? What about the authors of the Design Patterns Smalltalk Companion? James Robertson's Podcast was wonderful and I miss it a lot. But he had people on the podcast who also made it a joyful experience: Willow Lucas-Smith and David Buck come to mind. Anybody remember how much energy Randal Schwartz once put into Smalltalk marketing? I have only been in the Smalltalk world for the last 26 years, and everybody I am mentioning here has been working on the foundation that has been built for them a long time before that. Rebecca Wirfs-Brock? Martin Fowler? Eric Clayberg? Joseph Pelrine are just a few names that immediately com to my mind. And I am sure I am forgetting more important names or even never heard of them. We are standing on the shoulders of Giants. Some may even not be aware of what they did. Partly because they have moved on to other languages and communities. Likas Renggli, Julian Fitzell, Vassily Bykov, Michel Bany, Martin Kobetic, many of the Eclipse guys. The list is just too long. Joachim Am 26.07.21 um 06:34 schrieb Richard Sargent: > Tudor Girba and his colleagues at feenk need to be recognized, too. > > In many respects, they have pushed Smalltalk to do things that, I > think, the original designers would have heartily endorsed. > > On Sun, Jul 25, 2021, 18:44 <horrido.hobbies(a)gmail.com > <mailto:horrido.hobbies@gmail.com>> wrote: > > Thanks to everyone for some great suggestions. > > Iâm coming around to the notion that I asked the wrong question. > In the context of Smalltalkâs 50th anniversary, the reason I asked > the question is because I noted that in APLâs 50th anniversary > celebration, they gave an award to the one individual who was > deemed the greatest contributor to APL. So I thought I might do > the same for Smalltalk. > > But now, Iâm seeing that was somewhat wrongheaded. Yes, it is the > community that made Smalltalk what it was, and what it is today. > Singling out one individual seems unfair, and rather arbitrary. > > Maybe Iâll conduct a poll with these suggestions and let the > Smalltalk community express their support with their votes. That > may even provide some insight. > -- ----------------------------------------------------------------------- Objektfabrik Joachim Tuchel mailto:jtuchel@objektfabrik.de Fliederweg 1 http://www.objektfabrik.de D-71640 Ludwigsburg http://joachimtuchel.wordpress.com Telefon: +49 7141 56 10 86 0 Fax: +49 7141 56 10 86 1

July 26, 2021

Re: Zinc exception logging

by Sven Van Caekenberghe

Hi Esteban, > On 26 Jul 2021, at 02:47, Esteban Maringolo <emaringolo(a)gmail.com> wrote: > > Is there a way to have a "stack dump" response in Zinc? > > #debugMode is good for development, but I'm having an "Internal Error: > 4", that I don't know how to trace. > > * Trying 167.71.182.110... > * TCP_NODELAY set > * Connected to fore.base.golf (167.71.182.110) port 8090 (#0) >> GET /web HTTP/1.1 >> Host: x.y.z:8090 >> User-Agent: curl/7.58.0 >> Accept: */* >> > < HTTP/1.1 500 Internal Server Error > < Content-Length: 512 > < Content-Type: text/plain;charset=utf-8 > < Server: Zinc HTTP Components 1.0 (Pharo/8.0) > < Date: Mon, 26 Jul 2021 00:45:39 GMT > < > * Connection #0 to host x.y.z left intact > Internal Error: 4 > > Regards! > > Esteban A. Maringolo Searching the source code of my image, I only find 'Internal Error' in WAExceptionHandler/WAResponseGenerator, not in Zn code. Furthermore, printing 4 there would be weird (as it is the exception message). Is this a plain Zn response or a Seaside response ? I know that Seaside has more logging options. Zn does not have an exception logging mechanism that dumps a stack trace, only the debugMode for interactive development. I see you used curl, can't you fire the request against your dev image then ? Sven

July 26, 2021

Re: The Greatest Contributors to Smalltalk since 1980

by Richard Sargent

Tudor Girba and his colleagues at feenk need to be recognized, too. In many respects, they have pushed Smalltalk to do things that, I think, the original designers would have heartily endorsed. On Sun, Jul 25, 2021, 18:44 <horrido.hobbies(a)gmail.com> wrote: > Thanks to everyone for some great suggestions. > > Iâm coming around to the notion that I asked the wrong question. In the > context of Smalltalkâs 50th anniversary, the reason I asked the question is > because I noted that in APLâs 50th anniversary celebration, they gave an > award to the one individual who was deemed the greatest contributor to APL. > So I thought I might do the same for Smalltalk. > > But now, Iâm seeing that was somewhat wrongheaded. Yes, it is the > community that made Smalltalk what it was, and what it is today. Singling > out one individual seems unfair, and rather arbitrary. > > Maybe Iâll conduct a poll with these suggestions and let the Smalltalk > community express their support with their votes. That may even provide > some insight. >

July 26, 2021

Re: The Greatest Contributors to Smalltalk since 1980

by horrido.hobbies＠gmail.com

Thanks to everyone for some great suggestions. Iâm coming around to the notion that I asked the wrong question. In the context of Smalltalkâs 50th anniversary, the reason I asked the question is because I noted that in APLâs 50th anniversary celebration, they gave an award to the one individual who was deemed the greatest contributor to APL. So I thought I might do the same for Smalltalk. But now, Iâm seeing that was somewhat wrongheaded. Yes, it is the community that made Smalltalk what it was, and what it is today. Singling out one individual seems unfair, and rather arbitrary. Maybe Iâll conduct a poll with these suggestions and let the Smalltalk community express their support with their votes. That may even provide some insight.

July 26, 2021

Zinc exception logging

by Esteban Maringolo

Is there a way to have a "stack dump" response in Zinc? #debugMode is good for development, but I'm having an "Internal Error: 4", that I don't know how to trace. * Trying 167.71.182.110... * TCP_NODELAY set * Connected to fore.base.golf (167.71.182.110) port 8090 (#0) > GET /web HTTP/1.1 > Host: x.y.z:8090 > User-Agent: curl/7.58.0 > Accept: */* > < HTTP/1.1 500 Internal Server Error < Content-Length: 512 < Content-Type: text/plain;charset=utf-8 < Server: Zinc HTTP Components 1.0 (Pharo/8.0) < Date: Mon, 26 Jul 2021 00:45:39 GMT < * Connection #0 to host x.y.z left intact Internal Error: 4 Regards! Esteban A. Maringolo

July 26, 2021