- Pharo-users - lists.pharo.org

News entries wanted for Pharo website
by Marcus Denker Dec. 15, 2013

Dec. 15, 2013

Hi, We are posting news entries on Pharo.org (and twitter + google plus)â¦ often it is easy to transform an email announcement into a news entry, but sometimes it takes quite an effort. So if you want your news to be published, please send a short text + a picture or link to youtube for the pharo news feed. Itâs much easier to post something that needs no brain power than something where one needs to think how to announce it so it makes senseâ¦ So -> if you announce something via mail, send me (or to the list, or add an issue tacker entry)( a version to be put on pharo.org -> if you have something where you wonder why this was never put there, do the same ;-) Marcus

1 0

ConnectionClosed issues with Zinc
by Esteban A. Maringolo Dec. 14, 2013

Dec. 14, 2013

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

3 6

Pharo on Docker
by Esteban A. Maringolo Dec. 13, 2013

Dec. 13, 2013

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

2 1

Glorp PostgreSQL Drivers in Ubuntu 12.04
by Esteban A. Maringolo Dec. 13, 2013

Dec. 13, 2013

Hi there, I'm moving my development server (Windows) into production (Ubuntu 12.04) and I'm facing some issues with, among other things, the PostgreSQL driver. I'm getting an: "Error: External module not found" My database login is as follows: sampleLogin ^Login new database: (PostgreSQLPlatform new characterEncoding: 'utf8'); username: 'user'; password: 'secret'; connectString: 'localhost_db'; encodingStrategy: (DBXStaticEncoding newForEncoding: #utf8). And then just a regular accessor for the previous login: DatabaseAccessor forLogin: self sampleLogin. How should I install the dependencies? I have a libpq.dll in my windows environment, and I already installed libpq5 drivers in the server (apt-get install libpq5) What else should I do? Regards, Esteban A. Maringolo

7 15

MooseDay schedule
by Marcus Denker Dec. 13, 2013

Dec. 13, 2013

>> 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 >>

1 0

Is OpenDBX alive?
by Esteban A. Maringolo Dec. 13, 2013

Dec. 13, 2013

As a side question about my dependency issues... Is the OpenDBX project still active? Not its Smalltalk side, I mean... the core project. To me it seems like it was a great attempt to have something similar to ODBC but for all platforms. But I'm not seeing a wide usage of it in the wild. Seems like everybody else is happy with their JDBC, DBI and similar "DB layers" on the language side. Regards, Esteban A. Maringolo

4 4

Nginx Reverse Proxying
by Esteban A. Maringolo Dec. 12, 2013

Dec. 12, 2013

Hi there, I continue with my server setup... I'm trying to setup nginx as the HTTP front end. I have two servers, one for the Zinc SeasideAdaptor running on 8080 and another server running on 8081 (maybe there is a way to run both on the same server). ZnZincServerAdaptor startOn: 8080. ZnServer startDefaultOn: 8081. ZnServer default delegate: (ZnWebSocketDelegate map: 'ws' to: MyWsHandler new). In the Seaside side I have three "entry points": /app : which is a standard Seaside application /app-api/v1/ :which is a WARestfulHandler (REST API, with Basic HTTP Auth) /assets (File Directory for static files) I plan to put a nginx in front of those services, mainly for performance purpose of static file serving, and to avoid the exposure of my server directy to internet. Supposing the hostname is app01.mydomain.com ? My requirements are 1) I want that http://app01.mydomain.com/app be forwarded to http://localhost:8080/app, the same with app-api 2) Same for websockets, maybe to http://ws01.mydomain.com to http://localhost:8081/ws 3) http://app01.mydomain.com/assets be served statically. I know this is a request which isn't directly related with Pharo nor Seaside, but given the fact it's a common setup I hope somebody will help me with this. Sorry for the crosspost. Regards, Esteban A. Maringolo

1 0

[launcher] fixed layout of tollbar buttons
by btc＠openinworld.com Dec. 12, 2013

Dec. 12, 2013

Hi Damien, Benjamin.V.R. I see in "PharoLauncher-Spec-DamienCassou.17" that the layout of the toolbar buttons is fixed, which is really great. However in trying to learn from that for future reference, seeing it relies on an empty column being added "seems" like a hack to workaround some issue with Spec. Should that be logged as a bug? Or is there some greater architectural background for this that I am missing? PhLCommandTooldbar>>dynamicLayout "(newer)" "Create a spec layout with 1 row, and as many columns as there are command classes" ^ SpecLayout composed newRow: [ :r | r newColumn: [ :col | ]. "<--------FIX" self commandClasses do: [ :commandClass | r add: commandClass specId width: self class toolbarHeight ] ] I tried digging into Spec to look at this myself, and my naive intuition felt it was something to do with initialization of LayoutFrames, but that was my first exposure to LayoutFrame and I quickly got lost. However I was reminded of Igor commenting "i only know that if you using proportional layout, then if you want submorphs to be laid down properly, you must use addMorph:fullFrame: instead of just addMorph:" [1] I tried this with a few hacks, but tit was really shooting in the dark and unsuccessful. [1] https://www.mail-archive.com/pharo-dev@lists.pharo.org/msg08838.html cheers -ben

2 4

Re: [Pharo-users] [Pharo-dev] [Pharo Trick: #0009] 'Enable halt/inspect once' button
by btc＠openinworld.com Dec. 12, 2013

Dec. 12, 2013

1 0

[Pharo Trick: #0008] - If appropriate use the memory file system
by Torsten Bergmann Dec. 12, 2013

Dec. 12, 2013

----------------------------------------------------------------------------------- [Pharo Trick: #0008] - If appropriate use the memory file system ----------------------------------------------------------------------------------- Works in: Pharo 2.0, 3.0, ... ----------------------------------------------------------------------------------- It may be possible that your application has to export some data into an external file in a specific format. Pharo allows writing into a file using Streams as any other Smalltalk. But sometimes you want to check the result right from the Pharo image without having to look into the external file using external tools. So the trick here is to use the memory file system instead of the real disk file system during development since also no cleanup on the hard disk is required afterwards: |fs| fs := FileSystem memory. (fs / 'hello.txt') writeStreamDo: [:s| s nextPutAll: 'Exported data' ]. (fs / 'hello.txt') readStream contents. The memory filesystem is also nice if you need to work with files within a unit test. So when the test runs nothing remains left on the hard disk. If you need a real copy of the file on the hard disk you can also copy from the memory file system to the disk file system using #copyTo: |fs| fs := FileSystem memory. (fs / 'hello.txt') writeStreamDo: [:s| s nextPutAll: 'Memory first and then disk' ]. (fs / 'hello.txt') copyTo: (FileSystem disk / 'exactCopy.txt')

3 2