[Pharo-project] Why errors are not logged in the Transcript?
It should be fine to have stack traces into the Transcript as a default behavior in addition to the debugger. Guille :)
I disagree strongly. . . The transcript should maintain its purpose that is to show information 'intentionally' written to by the programmer. If we feel the debugger info should be saved in some other place we should consider an alternate way of saving it, including, perhaps a log file and a {radio}button in the debugger itself. My 0.19999999..... Em 21/04/2011 15:19, Guillermo Polito < guillermopolito@gmail.com > escreveu: It should be fine to have stack traces into the Transcript as a default behavior in addition to the debugger. Guille :)
Maybe I'm to viced by other techs, but reading and sharing the stack trace is a common task in everyday work. And I think the transcript as a "log console" and useful only for debugging purposes.... And putting info into files is not the smalltalk way... BTW, I hate searching for the stacktraces into the PharoDebug.log :P On Fri, Apr 22, 2011 at 1:24 AM, <csrabak@bol.com.br> wrote:
I disagree strongly. . . The transcript should maintain its purpose that is to show information 'intentionally' written to by the programmer.
If we feel the debugger info should be saved in some other place we should consider an alternate way of saving it, including, perhaps a log file and a {radio}button in the debugger itself.
My 0.19999999.....
Em 21/04/2011 15:19, Guillermo Polito < guillermopolito@gmail.com > escreveu: It should be fine to have stack traces into the Transcript as a default behavior in addition to the debugger.
Guille :)
Yes, reading and sharing the stack trace is a common task, but to do that you don't need to pollute the Transcript. If you want to do a (albeit crude) parallel to other realms, think of Transcript as the Unix stdout and the debug pane as the stderr. Putting *useful* info into files so others than the programmer can consume it's not contrary to Smalltalk. As parallel, Squeak mailing lists include all the diffs for changes in the e-mail's bodies, which can be considered as the same as filing them out. Maybe you hating on searching PharoDebug.log stems from lack of an appropriate tool? Em 22/04/2011 10:28, Guillermo Polito < guillermopolito@gmail.com > escreveu: Maybe I'm to viced by other techs, but reading and sharing the stack trace is a common task in everyday work. And I think the transcript as a "log console" and useful only for debugging purposes.... And putting info into files is not the smalltalk way... BTW, I hate searching for the stacktraces into the PharoDebug.log :P On Fri, Apr 22, 2011 at 1:24 AM, <csrabak@bol.com.br> wrote: I disagree strongly. . . The transcript should maintain its purpose that is to show information 'intentionally' written to by the programmer. If we feel the debugger info should be saved in some other place we should consider an alternate way of saving it, including, perhaps a log file and a {radio}button in the debugger itself. My 0.19999999..... Em 21/04/2011 15:19, Guillermo Polito < guillermopolito@gmail.com > escreveu: It should be fine to have stack traces into the Transcript as a default behavior in addition to the debugger. Guille :)
On 22 April 2011 06:24, <csrabak@bol.com.br> wrote:
I disagree strongly. . . The transcript should maintain its purpose that is to show information 'intentionally' written to by the programmer.
and so, what if my intent is to show error stack trace on transcript? in headless mode i would prefer to log everything to some default logging facility.. guess which one? yes, it is Transcript. But then also i may want to use different stream for transcript in headless mode (to redirect output to stdout or file), but that won't require changing the error logging, which could simply output everything to Transcript.
If we feel the debugger info should be saved in some other place we should consider an alternate way of saving it, including, perhaps a log file and a {radio}button in the debugger itself.
My 0.19999999.....
Em 21/04/2011 15:19, Guillermo Polito < guillermopolito@gmail.com > escreveu: It should be fine to have stack traces into the Transcript as a default behavior in addition to the debugger.
Guille :)
-- Best regards, Igor Stasenko AKA sig.
Em 22/04/2011 11:10, Igor Stasenko < siguctua@gmail.com > escreveu:
On 22 April 2011 06:24, <csrabak@bol.com.br> wrote:
I disagree strongly. . . The transcript should maintain its purpose that is to show information 'intentionally' written to by the programmer. and so, what if my intent is to show error stack trace on transcript?
You as programmer can do what you want in your system. However having this arrangement as default for all users/programmers is IMNSO *wrong*.
in headless mode i would prefer to log everything to some default logging facility.. guess which one? yes, it is Transcript.
In a headless mode, probably the only user would be a sysadmin, so the choice looks sensible, although I still think that the stdout/stderr metaphor is more appropriate.
But then also i may want to use different stream for transcript in headless mode (to redirect output to stdout or file), but that won't require changing the error logging, which could simply output everything to Transcript.
If all these choices are not hardwired in the image, or at least made flexible enough so the defaults can be easily changed, I think it's OK. What I'm opposing is having this as standard in Pharo. Regards, -- Cesar Rabak
participants (3)
-
csrabak@bol.com.br -
Guillermo Polito -
Igor Stasenko