- Pharo-dev - lists.pharo.org

Re: [Pharo-project] Primitives

by Marcus Denker

On May 8, 2012, at 5:00 AM, Santiago Bragagnolo wrote: > Hi all, yes. me again. > > I'm searching for a primitive-definition, something like > code : what.it.do : relatedType if apply. > > Is there something like that? > No, sadly not. We need a table that has infos on input, side-effects, if the primitive is there just for speed (fall back code implements it, too)... things like that. And, of course, we should really clean up primitives and remove the ones not needed. (same for Plugins). Marcus -- Marcus Denker -- http://marcusdenker.de

May 8, 2012

Re: [Pharo-project] Jenkins is unreachable right now

by Marcus Denker

On May 8, 2012, at 9:34 AM, Pavel Krivanek wrote: > Jenkins is unreachable right now :-( > > -- Pavel works for me. -- Marcus Denker -- http://marcusdenker.de

May 8, 2012

Re: [Pharo-project] ServerDirectory

by Marcus Denker

On May 7, 2012, at 12:58 PM, Marcus Denker wrote: > > On May 7, 2012, at 12:37 PM, Sven Van Caekenberghe wrote: > >> >> On 07 May 2012, at 11:48, Camillo Bruni wrote: >> >>> how much do we still need ServerDirectory in 2.0? >>> Isn't that almost fully included by Zn? >>> >>> ServerDirectory features a lot of references to FileDirectory hence I would like to nuke it :D >> >> I also don't really understand why ServerDirectory exists. >> MCConfigurationBrowser seems to use it. >> >> If ServerDirectory does something that cannot be done with ZnClient, I am all ears. >> > > ServerDirectory needs to be removed. It was just a mapping of names to servesr so that > one could use symbolic names for pre-defined servers. Quite strange. > The last client of ServerDirectory is FtpUrl... and the method there looks very hackish and can't even be compiled. The code in ServerDirectory is very bad... there is no value in that. As it is perfectly possible to just use FTPClient directly (see Monticello), I think we should just remove ServerDirectory and FtpUrl FtpUrl is not really useful anyway... ftp is using non-encrypted passwords, so everyone using it is doing somethign wrong anyway. A good FTP client libeary in the spirit of Zinc would be nice of course in the long run. http://code.google.com/p/pharo/issues/detail?id=3639 Anyone against? -- Marcus Denker -- http://marcusdenker.de

May 8, 2012

[Pharo-project] Jenkins is unreachable right now

by Pavel Krivanek

Jenkins is unreachable right now :-( -- Pavel

May 8, 2012

Re: [Pharo-project] Primitives

by Frank Shearar

Hi Santiago, On 8 May 2012 03:59, Santiago Bragagnolo <santiagobragagnolo(a)gmail.com> wrote: > Hi all, yes. me again. > > Â I'm searching for a primitive-definition, something like > code : what.it.do : relatedType if apply. > > Is there something like that? Are you looking to type a primitive call? frank > Thanks in advance. > > Santiago.

May 8, 2012

Re: [Pharo-project] Running in background with OSProcess

by Sean P. DeNigris

David T. Lewis wrote > > ProxyPipeline command: '/bin/sleep 10' > This gives you exactly what you expect to see. > Okay, it worked! However, I'm confused by the string escaping. Because I'm on a Mac, and only later discovered that meant I'm on Unix, I have lots of paths with spaces and even [gasp] apostrophes in them (the second makes Smalltalk strings a lot of fun). One thing that I loved about OSProcess is that I felt like I had more control over strings (e.g. paths) in commands because I could construct my paths in Smalltalk and then just double-quote the final product. However, using ProxyPipeline, it seems that CommandShell has layered on its own escaping rules that are different from Bash's. So, although I can write: p := PipeableOSProcess command: '/usr/bin/java -DJENKINS_HOME="/path/with/a space/and/an/apostrophe/isn''t/fun/Pharo-1.4/.jenkins"...' with ProxyPipeline, I apparently have to write: p2 := ProxyPipeline command: '/usr/bin/java -DJENKINS_HOME=/path/with/a\ space/and/an/apostrophe/isn\''t/fun/Pharo-1.4/.jenkins...' I.e. "enclose in double-quotes" vs. "escape single-quotes and spaces with preceding backslash" Can I somehow circumvent these extra rules so I don't have to adapt all my commands? What is their purpose? Getting closer and closer to having this work... Thanks. Sean -- View this message in context: http://forum.world.st/Running-in-background-with-OSProcess-tp4615534p461658… Sent from the Pharo Smalltalk mailing list archive at Nabble.com.

May 8, 2012

[Pharo-project] Nested progress bars

by Martin Dias

Hi all. Given the next code, could you explain me why nested progress bars are not handled by my #on:do:? How can I achieve that? [ 'main task' displayProgressFrom: 0 to: 5 during: [ :bar | 0 to: 5 do: [:x | bar value: x. (Delay forMilliseconds: 200) wait. 'nested task ' , x printString displayProgressFrom: 5 to: 10 during: [ :bar2 | 5 to: 10 do: [:x2 | bar2 value: x2. (Delay forMilliseconds: 200) wait] ] ] ] ] on: ProgressInitiationException do: [ :ex | ex sendNotificationsTo: [ :min :max :curr | self traceCr: curr ] ]. Thanks in advance. MartÃn

May 8, 2012

[Pharo-project] Primitives

by Santiago Bragagnolo

Hi all, yes. me again. I'm searching for a primitive-definition, something like code : what.it.do : relatedType if apply. Is there something like that? Thanks in advance. Santiago.

May 8, 2012

Re: [Pharo-project] Nautilus load problem

by Sean P. DeNigris

Marcus Denker-4 wrote > > Nautilus... getting it ready will be double as hard if it has to run on > both 1.4 and 2.0. > Is Nautilus unsupported on 1.4? There have been many suggestions to load it in preference to the default browser in 1.4, so we should be clear whether it will be expected to work. I guess then OB would be the only workable option... but Nautilus is so cool - it's hard to go back, lol!! -- View this message in context: http://forum.world.st/Nautilus-load-problem-tp4603363p4616242.html Sent from the Pharo Smalltalk mailing list archive at Nabble.com.

May 8, 2012

Re: [Pharo-project] Running in background with OSProcess

by David T. Lewis

On Mon, May 07, 2012 at 11:46:30PM +0200, Norbert Hartl wrote: > > Am 07.05.2012 um 20:52 schrieb Sean P. DeNigris: > > > How would I run '/usr/bin/java -jar "/path/to/jenkins.war" &' with OSProcess? > > Specifically I want to know the process number of the java child process. > > > > I unsuccessfully tried: > > * PipeableOSProcess command: '/usr/bin/java -jar "/path/to/jenkins.war" &', > > which successfully launches java, but I don't know its pid because the > > returned process is the completed parent "sh" that launched it > > * PipeableOSProcess command: '/usr/bin/java -jar "/path/to/jenkins.war"' > > (without ampersand), which also launches successfully, and leaves the parent > > "sh" process running as well (not as clean), but I still don't know the java > > pid > > > If this is all shell handling then you should be able to exec your command. An > > exec /usr/bin/java -jar "/path/to/jenkins.war" > > (if it works) will replace the shell with the execution of the java process. Therefor the pid is the pid of the java process. But Yannis approach looks a lot better from here :) > That's an excellent suggestion and it will do exactly what Sean is looking to do. But it may a bit confusing because the process proxy will think that it is running a shell even though that shell has done an exec to replace itself with the /usr/bin/java executable. So I'll try to add some more explanation here, using the /bin/sleep program as an example rather that /usr/bin/java. For this example, the "long running process" is /bin/sleep running for ten seconds, so /bin/sleep takes the place of /usr/bin/java for the example. Here is what Sean was originally doing: PipeableOSProcess command: '/bin/sleep 10' This uses a unix shell to run the command, and lets the shell do the argument parsing (to pass the argument "10" to the /bin/sleep program). The problem then is that the actual /bin/sleep program is forked by the unix shell, so you have no direct access to that subprocess from your image, and you do not know the pid of the process that is actually running your program. By adding the exec shell command, you get similar results except that the /bin/sleep program replaces /bin/sh (or bash or whatever the shell was), such that the pid of the process running /bin/sleep remains the same as that for /bin/sh (because it is actually the same process). PipeableOSProcess command: 'exec /bin/sleep 10' So this does what is wanted. However, it might be confusing when you inspect the process proxy in your image, because your image still thinks that a shell is being run, even though the /bin/sh executable has now been replaced by /bin/sleep running in that same external process. When you inspect the process proxy, it will show the program name '/bin/sh' even though that executable program has now been replaced by /bin/sleep as a result of the exec. One way to avoid this confusion is to do the unix shell processing in Smalltalk rather than relying on an external unix shell. That is what CommandShell does, so you could run the /bin/sleep program like this and let CommandShell do all the work that would otherwise be done by /bin/sh. CommandShell command: '/bin/sleep 10' But this is still not quite right, because CommandShell will also open a user interface window, and that is not what you want. CommandShell itself uses a ProxyPipeline to represent one or more process proxies from a command line, so you can use that class directly to evaluate a command line: ProxyPipeline command: '/bin/sleep 10' This gives you exactly what you expect to see. The command line will be parsed in a manner similar to a unix shell, and it will set up a single PipeableOSProcess that runs the /bin/sleep program, with a process proxy that gives both the pid of the process for /bin/sleep, and the actual program name of the /bin/sleep program that is being run. HTH, Dave

May 8, 2012