December 2013 - Pharo-users

Pharo on Docker

by Esteban A. Maringolo

Branch from thread: "ConnectionClosed issues with Zinc" 2013/12/13 Sven Van Caekenberghe <sven(a)stfx.eu>: > On 13 Dec 2013, at 05:01, Esteban A. Maringolo <emaringolo(a)gmail.com> wrote: >> The second time everything went smoother and faster. I'm thinking >> about setting up a Docker [3] container for Pharo, and then use it in >> Digital Ocean as well [4]. > > Is docker not 64-bit ? I am not sure running a 32-bit Pharo VM inside it will be easy, but maybe it is. Can't Pharo run on 64bit host OSes if you install ia32-libs? The VM will sill be 32 bits, with its known limitations. But the idea of setting up the environment locally and then deploying the same environment to a production server sounds really interesting. "Containerization is the new virtualization" they say. Regards! Esteban A. Maringolo

Dec. 13, 2013

Re: [Pharo-users] Glorp PostgreSQL Drivers in Ubuntu 12.04

by Esteban A. Maringolo

2013/12/13 Pierce Ng <pierce(a)samadhiweb.com>: > On Thu, Dec 12, 2013 at 12:53:51AM -0200, Esteban A. Maringolo wrote: > I recommended SpsSplitPasswordStore: > > http://samadhiweb.com/blog/2013.08.11.splitpasswordstore.html Thank for your suggestion. Even when it's security through obscurity I like the approach. However, the pass is not stored in the application. Also, the database server only accepts connections from localhost. Host to which I can only connect by means of RSA keys (no password). So it's a risk I assume for convenience. :) Regards! Esteban A. Maringolo

Dec. 13, 2013

Re: [Pharo-users] ConnectionClosed issues with Zinc

by StÃ©phane Ducasse

esteban it would be really great if you could extract from your experience a kind of blog, book chapter whatever to share it with other developers around the world. Stef On Dec 13, 2013, at 1:36 AM, Esteban A. Maringolo <emaringolo(a)gmail.com> wrote: > I'm having issues with a connection coming from an Android app > connecting both trough an nginx proxy, but also using some REST > testing tools: > > nginx error log: > 2013/12/12 23:23:17 [error] 701#0: *49 upstream prematurely closed > connection while reading response header from upstream, client: > XX.XX.XX.XX, server: app01.mydomain.com, request: "POST > /app-api/v1/auth/login HTTP/1.1", upstream: > "http://127.0.0.1:8080/app-api/v1/auth/login", host: > "app01.mydomain.com:80" > (END) > > ZnZincSeasideAdapter log: > 2013-12-12 23:23:17 897330 D Executing request/response loop > 2013-12-12 23:23:17 897330 I Read a ZnRequest(POST /app-api/v1/auth/login) > 2013-12-12 23:23:17 897330 T POST /app-api/v1/auth/login 200 114B 42ms > 2013-12-12 23:23:17 897330 D ConnectionClosed: Cannot write data while > writing response > 2013-12-12 23:23:17 897330 D Closing stream > > I'm using Basic HTTP Auth in each request. > > But what's weird is that with curl it works perfectly fine. > > Again, I think this is not strictly Zinc related, but it doesn't work > connecting directly to the Zinc server either: > > 2013-12-13 00:17:02 939035 D Executing request/response loop > 2013-12-13 00:17:02 939035 I Read a ZnRequest(POST /app-api/v1/auth/login) > 2013-12-13 00:17:02 939035 T POST /app-api/v1/auth/login 200 114B 12ms > 2013-12-13 00:17:02 939035 D ConnectionClosed: Cannot write data while > writing response > 2013-12-13 00:17:02 939035 D Closing stream > > I don't know if this has to do with the wireless provider doing some > proxying behind the wall (which they do), but given the fact that if I > try using some REST testing tools from my computer and they also fail, > I think it has to do with something else. > > At nginx I'm forwarding the Authorization header to the Zinc Server: > > proxy_set_header Authorization $http_authorization; > proxy_pass_header Authorization; > > Is there a way I can log the complete request headers at the Zinc > side, and if possible also at nginx? > > Regards, > > > Esteban A. Maringolo >

Dec. 13, 2013

MooseDay schedule

by Marcus Denker

>> De : Anne Etien <anne.etien(a)univ-lille1.fr> >> >> Hi, >> >> Please find bellow the schedule of the MooseDay >> http://www.lifl.fr/~etien/MooseDay.html >> The idea is really to present works developed on top of (or around) Moose and to have fruitful discussions. So I planned 30 mn per topic, but the schedule is quite flexible. Lot of time is foreseen for discussions. >> >> 10h: welcome >> 10h30: APIEvolutionMiner - Andre Hora >> 11h: FamixDiff - Nicolas Anquetil >> 11h30: Orion - Jannik Laval and Anne Etien >> >> 12h: Lunch offered by ESUG. >> >> 14h: First works on Version - Yuriy Tymchuk >> 14h30: Roassal - Alexandre Bergel >> >> 15h-15h30 break >> >> 15h30: GraphET - Alexandre Bergel >> 16h: Use of GraphET - Serge Stinckwich >> 16h30: Around tests - Alain Plantec >> >> Reminder, the MooseDay will take place in Inria B building room B31. Directions to come can be found there (http://www.inria.fr/en/centre/lille/overview/locations) >> >> See you next week in Lille ;o) >> >> Anne >>

Dec. 13, 2013

Re: [Pharo-users] ConnectionClosed issues with Zinc

by Sven Van Caekenberghe

On 13 Dec 2013, at 05:01, Esteban A. Maringolo <emaringolo(a)gmail.com> wrote: > Answering myself again (I knowâ¦). Thanks for posting all this, it is useful. > There is some issue in how the Basic Authentication is handled in > certain Android devices [1], which by chance (fortunately) is the same > I use to develop with, so I had to change the way the header was set > in the request, which basically I setup manually [2] > > So, only good news. The server is up and running. > In fact it is the second server, I created a second Droplet at Digital > Ocean because the first had Ubuntu 12.04 LTS and I wanted a newer > release to have the newer version of nignx as well. So Ubuntu 13.10 > 32bit it was. > > The second time everything went smoother and faster. I'm thinking > about setting up a Docker [3] container for Pharo, and then use it in > Digital Ocean as well [4]. Is docker not 64-bit ? I am not sure running a 32-bit Pharo VM inside it will be easy, but maybe it is. > It was a long day, but worth it. > > Best regards, > > Esteban A. Maringolo > > > [1] https://github.com/loopj/android-async-http/issues/262 > [2] http://stackoverflow.com/a/20558752/644918 > [3] http://www.docker.io (anybody using it?) > [4] https://www.digitalocean.com/community/articles/how-to-use-the-digitalocean… > > 2013/12/12 Esteban A. Maringolo <emaringolo(a)gmail.com>: >> Good news, bad news. :) >> >> I decided to test using another great tool bundled with Pharo itself: ZnClient. >> >> >> "Going through nginx" >> ZnClient new >> url: 'http://myserver.com/app-api/v1/auth/login'; >> setBasicAuthenticationUsername: '24603458' password: '1234'; >> logToTranscript; >> post; >> response. >> >> And... as expected... it works! >> >> The same if connect directly to Zinc (:8080). >> >> So there is something wrong in the rest of the tools around... or I >> really don't know. >> >> But the HTTP client library we're using at the Android (Java) side is >> one of the most used (http://loopj.com/android-async-http/) but also >> fails with org.apache.http client. So... it's not a matter of client. >> >> I just need to understand the reason for the "Closed Connection". >> >> >> >> >> Esteban A. Maringolo >> >> >> 2013/12/12 Esteban A. Maringolo <emaringolo(a)gmail.com>: >>> I'm having issues with a connection coming from an Android app >>> connecting both trough an nginx proxy, but also using some REST >>> testing tools: >>> >>> nginx error log: >>> 2013/12/12 23:23:17 [error] 701#0: *49 upstream prematurely closed >>> connection while reading response header from upstream, client: >>> XX.XX.XX.XX, server: app01.mydomain.com, request: "POST >>> /app-api/v1/auth/login HTTP/1.1", upstream: >>> "http://127.0.0.1:8080/app-api/v1/auth/login", host: >>> "app01.mydomain.com:80" >>> (END) >>> >>> ZnZincSeasideAdapter log: >>> 2013-12-12 23:23:17 897330 D Executing request/response loop >>> 2013-12-12 23:23:17 897330 I Read a ZnRequest(POST /app-api/v1/auth/login) >>> 2013-12-12 23:23:17 897330 T POST /app-api/v1/auth/login 200 114B 42ms >>> 2013-12-12 23:23:17 897330 D ConnectionClosed: Cannot write data while >>> writing response >>> 2013-12-12 23:23:17 897330 D Closing stream >>> >>> I'm using Basic HTTP Auth in each request. >>> >>> But what's weird is that with curl it works perfectly fine. >>> >>> Again, I think this is not strictly Zinc related, but it doesn't work >>> connecting directly to the Zinc server either: >>> >>> 2013-12-13 00:17:02 939035 D Executing request/response loop >>> 2013-12-13 00:17:02 939035 I Read a ZnRequest(POST /app-api/v1/auth/login) >>> 2013-12-13 00:17:02 939035 T POST /app-api/v1/auth/login 200 114B 12ms >>> 2013-12-13 00:17:02 939035 D ConnectionClosed: Cannot write data while >>> writing response >>> 2013-12-13 00:17:02 939035 D Closing stream >>> >>> I don't know if this has to do with the wireless provider doing some >>> proxying behind the wall (which they do), but given the fact that if I >>> try using some REST testing tools from my computer and they also fail, >>> I think it has to do with something else. >>> >>> At nginx I'm forwarding the Authorization header to the Zinc Server: >>> >>> proxy_set_header Authorization $http_authorization; >>> proxy_pass_header Authorization; >>> >>> Is there a way I can log the complete request headers at the Zinc >>> side, and if possible also at nginx? >>> >>> Regards, >>> >>> >>> Esteban A. Maringolo >

Dec. 13, 2013

Re: [Pharo-users] Glorp PostgreSQL Drivers in Ubuntu 12.04

by Pierce Ng

On Thu, Dec 12, 2013 at 12:53:51AM -0200, Esteban A. Maringolo wrote: > sampleLogin > ^Login new > database: (PostgreSQLPlatform new characterEncoding: 'utf8'); > username: 'user'; > password: 'secret'; <shamelessplug> I recommended SpsSplitPasswordStore: http://samadhiweb.com/blog/2013.08.11.splitpasswordstore.html ^Login new database: (PostgreSQLPlatform new characterEncoding: 'utf8'); username: 'user'; password: (SpsSplitPasswordStore readFrom: 'user.dat'); ... </shamelessplug> I'm thinking to remove the code that takes random strings from /dev/random and do it in a more portable way without weakening security too much. -- Pierce Ng http://samadhiweb.com/blog/

Dec. 13, 2013

Re: [Pharo-users] ConnectionClosed issues with Zinc

by Esteban A. Maringolo

Answering myself again (I know...). There is some issue in how the Basic Authentication is handled in certain Android devices [1], which by chance (fortunately) is the same I use to develop with, so I had to change the way the header was set in the request, which basically I setup manually [2] So, only good news. The server is up and running. In fact it is the second server, I created a second Droplet at Digital Ocean because the first had Ubuntu 12.04 LTS and I wanted a newer release to have the newer version of nignx as well. So Ubuntu 13.10 32bit it was. The second time everything went smoother and faster. I'm thinking about setting up a Docker [3] container for Pharo, and then use it in Digital Ocean as well [4]. It was a long day, but worth it. Best regards, Esteban A. Maringolo [1] https://github.com/loopj/android-async-http/issues/262 [2] http://stackoverflow.com/a/20558752/644918 [3] http://www.docker.io (anybody using it?) [4] https://www.digitalocean.com/community/articles/how-to-use-the-digitalocean… 2013/12/12 Esteban A. Maringolo <emaringolo(a)gmail.com>: > Good news, bad news. :) > > I decided to test using another great tool bundled with Pharo itself: ZnClient. > > > "Going through nginx" > ZnClient new > url: 'http://myserver.com/app-api/v1/auth/login'; > setBasicAuthenticationUsername: '24603458' password: '1234'; > logToTranscript; > post; > response. > > And... as expected... it works! > > The same if connect directly to Zinc (:8080). > > So there is something wrong in the rest of the tools around... or I > really don't know. > > But the HTTP client library we're using at the Android (Java) side is > one of the most used (http://loopj.com/android-async-http/) but also > fails with org.apache.http client. So... it's not a matter of client. > > I just need to understand the reason for the "Closed Connection". > > > > > Esteban A. Maringolo > > > 2013/12/12 Esteban A. Maringolo <emaringolo(a)gmail.com>: >> I'm having issues with a connection coming from an Android app >> connecting both trough an nginx proxy, but also using some REST >> testing tools: >> >> nginx error log: >> 2013/12/12 23:23:17 [error] 701#0: *49 upstream prematurely closed >> connection while reading response header from upstream, client: >> XX.XX.XX.XX, server: app01.mydomain.com, request: "POST >> /app-api/v1/auth/login HTTP/1.1", upstream: >> "http://127.0.0.1:8080/app-api/v1/auth/login", host: >> "app01.mydomain.com:80" >> (END) >> >> ZnZincSeasideAdapter log: >> 2013-12-12 23:23:17 897330 D Executing request/response loop >> 2013-12-12 23:23:17 897330 I Read a ZnRequest(POST /app-api/v1/auth/login) >> 2013-12-12 23:23:17 897330 T POST /app-api/v1/auth/login 200 114B 42ms >> 2013-12-12 23:23:17 897330 D ConnectionClosed: Cannot write data while >> writing response >> 2013-12-12 23:23:17 897330 D Closing stream >> >> I'm using Basic HTTP Auth in each request. >> >> But what's weird is that with curl it works perfectly fine. >> >> Again, I think this is not strictly Zinc related, but it doesn't work >> connecting directly to the Zinc server either: >> >> 2013-12-13 00:17:02 939035 D Executing request/response loop >> 2013-12-13 00:17:02 939035 I Read a ZnRequest(POST /app-api/v1/auth/login) >> 2013-12-13 00:17:02 939035 T POST /app-api/v1/auth/login 200 114B 12ms >> 2013-12-13 00:17:02 939035 D ConnectionClosed: Cannot write data while >> writing response >> 2013-12-13 00:17:02 939035 D Closing stream >> >> I don't know if this has to do with the wireless provider doing some >> proxying behind the wall (which they do), but given the fact that if I >> try using some REST testing tools from my computer and they also fail, >> I think it has to do with something else. >> >> At nginx I'm forwarding the Authorization header to the Zinc Server: >> >> proxy_set_header Authorization $http_authorization; >> proxy_pass_header Authorization; >> >> Is there a way I can log the complete request headers at the Zinc >> side, and if possible also at nginx? >> >> Regards, >> >> >> Esteban A. Maringolo

Dec. 13, 2013

Re: [Pharo-users] ConnectionClosed issues with Zinc

by Esteban A. Maringolo

Good news, bad news. :) I decided to test using another great tool bundled with Pharo itself: ZnClient. "Going through nginx" ZnClient new url: 'http://myserver.com/app-api/v1/auth/login'; setBasicAuthenticationUsername: '24603458' password: '1234'; logToTranscript; post; response. And... as expected... it works! The same if connect directly to Zinc (:8080). So there is something wrong in the rest of the tools around... or I really don't know. But the HTTP client library we're using at the Android (Java) side is one of the most used (http://loopj.com/android-async-http/) but also fails with org.apache.http client. So... it's not a matter of client. I just need to understand the reason for the "Closed Connection". Esteban A. Maringolo 2013/12/12 Esteban A. Maringolo <emaringolo(a)gmail.com>: > I'm having issues with a connection coming from an Android app > connecting both trough an nginx proxy, but also using some REST > testing tools: > > nginx error log: > 2013/12/12 23:23:17 [error] 701#0: *49 upstream prematurely closed > connection while reading response header from upstream, client: > XX.XX.XX.XX, server: app01.mydomain.com, request: "POST > /app-api/v1/auth/login HTTP/1.1", upstream: > "http://127.0.0.1:8080/app-api/v1/auth/login", host: > "app01.mydomain.com:80" > (END) > > ZnZincSeasideAdapter log: > 2013-12-12 23:23:17 897330 D Executing request/response loop > 2013-12-12 23:23:17 897330 I Read a ZnRequest(POST /app-api/v1/auth/login) > 2013-12-12 23:23:17 897330 T POST /app-api/v1/auth/login 200 114B 42ms > 2013-12-12 23:23:17 897330 D ConnectionClosed: Cannot write data while > writing response > 2013-12-12 23:23:17 897330 D Closing stream > > I'm using Basic HTTP Auth in each request. > > But what's weird is that with curl it works perfectly fine. > > Again, I think this is not strictly Zinc related, but it doesn't work > connecting directly to the Zinc server either: > > 2013-12-13 00:17:02 939035 D Executing request/response loop > 2013-12-13 00:17:02 939035 I Read a ZnRequest(POST /app-api/v1/auth/login) > 2013-12-13 00:17:02 939035 T POST /app-api/v1/auth/login 200 114B 12ms > 2013-12-13 00:17:02 939035 D ConnectionClosed: Cannot write data while > writing response > 2013-12-13 00:17:02 939035 D Closing stream > > I don't know if this has to do with the wireless provider doing some > proxying behind the wall (which they do), but given the fact that if I > try using some REST testing tools from my computer and they also fail, > I think it has to do with something else. > > At nginx I'm forwarding the Authorization header to the Zinc Server: > > proxy_set_header Authorization $http_authorization; > proxy_pass_header Authorization; > > Is there a way I can log the complete request headers at the Zinc > side, and if possible also at nginx? > > Regards, > > > Esteban A. Maringolo

Dec. 13, 2013

Re: [Pharo-users] Is OpenDBX alive?

by Pierce Ng

On Thu, Dec 12, 2013 at 07:01:07PM -0200, Esteban A. Maringolo wrote: > I don't know how good is to have a pure Smalltalk driver either, not > in terms of maintenance*, but in terms of performance, memory, etc. Linking to C libraries using FFI brings its own performance considerations with respect to blocking the VM. Speaking of which, what is the state of the art? Does the Cog+NB combo allow, say, to write an SQLite user-defined function in Smalltalk that is invoked by the C library per row of result? -- Pierce Ng http://samadhiweb.com/blog/

Dec. 13, 2013

ConnectionClosed issues with Zinc

by Esteban A. Maringolo

I'm having issues with a connection coming from an Android app connecting both trough an nginx proxy, but also using some REST testing tools: nginx error log: 2013/12/12 23:23:17 [error] 701#0: *49 upstream prematurely closed connection while reading response header from upstream, client: XX.XX.XX.XX, server: app01.mydomain.com, request: "POST /app-api/v1/auth/login HTTP/1.1", upstream: "http://127.0.0.1:8080/app-api/v1/auth/login", host: "app01.mydomain.com:80" (END) ZnZincSeasideAdapter log: 2013-12-12 23:23:17 897330 D Executing request/response loop 2013-12-12 23:23:17 897330 I Read a ZnRequest(POST /app-api/v1/auth/login) 2013-12-12 23:23:17 897330 T POST /app-api/v1/auth/login 200 114B 42ms 2013-12-12 23:23:17 897330 D ConnectionClosed: Cannot write data while writing response 2013-12-12 23:23:17 897330 D Closing stream I'm using Basic HTTP Auth in each request. But what's weird is that with curl it works perfectly fine. Again, I think this is not strictly Zinc related, but it doesn't work connecting directly to the Zinc server either: 2013-12-13 00:17:02 939035 D Executing request/response loop 2013-12-13 00:17:02 939035 I Read a ZnRequest(POST /app-api/v1/auth/login) 2013-12-13 00:17:02 939035 T POST /app-api/v1/auth/login 200 114B 12ms 2013-12-13 00:17:02 939035 D ConnectionClosed: Cannot write data while writing response 2013-12-13 00:17:02 939035 D Closing stream I don't know if this has to do with the wireless provider doing some proxying behind the wall (which they do), but given the fact that if I try using some REST testing tools from my computer and they also fail, I think it has to do with something else. At nginx I'm forwarding the Authorization header to the Zinc Server: proxy_set_header Authorization $http_authorization; proxy_pass_header Authorization; Is there a way I can log the complete request headers at the Zinc side, and if possible also at nginx? Regards, Esteban A. Maringolo

Dec. 13, 2013