Strings are fine when you don�t need specific behavior oder structure of your log messages anymore. A conversion to string should in most cases happen as late as possible. A log entry can still serve other purposes than represent an custom formatted string thing. Think about making all the exception handling in away that it can only signal strings. The same you feel now you would lose here is similar to the gain you�ll get making log messages objects ;) It is just that you are used to objects when it comes to exceptions and you are used to strings when comes to log messages. The former being a good idea the latter not.What's wrong with Strings anyway? But yeah, I understand your point.
http://smalltalkhub.com/#!/~StephaneDucasse/SystemLoggerWhat I have done for my own stuff is to have a TLog aspect with info, debug, error, warn and, well, Strings. But double dispatch could take care of that, like printOn:, storeOn: etc.
Do you have some code that I can look at?
anObject logOn: Logger default info.TLog>>info: anObject
PhilOn Tue, Dec 3, 2013 at 11:27 AM, St�phane Ducasse <stephane.ducasse@inria.fr> wrote:On Dec 3, 2013, at 10:34 AM, phil@highoctane.be wrote:I am working on this one.Don't kill the transcript, it is useful when working without a console.I will not kill it but it will just be dull outputer for my logger. In fact I'm missing a user interface (the idea was to build it with fernando) and right nowthe transcript just displays the printstring of my log objects. So I can even control what I want to output.more important I can filter the log ojbjects. For example I need to know when my framework finish to deal with a class and now this is mix with the compilerloghereaNow, adding the features you mention is not hard. And the top right triangle menu of Transcript could get a clue from the Workspace one with a Save As entry.
No i just DO NOT WANT STRINGS!!!!I once had an issue with things like (out of the top of my head) FileStream stdout nextPutAll: 'hello'; cr; flush. when I went headless. The image worked at one point and the froze/locked. Until this is sorted out in full, keep a transcript handy.PhilI gave the image to Guille who sent me back a version "unlocked" with Oz, but it still dies on Windows for example. Weird.---Philippe BackDramatic Performance ImprovementsMob: +32(0) 478 650 140 | Fax: +32 (0) 70 408 027Blog: http://philippeback.be | Twitter: @philippebackHigh Octane SPRLrue cour Boisacq 101 | 1301 Bierges | BelgiumPharo Consortium Member - http://consortium.pharo.org/Featured on the Software Process and Measurement Cast - http://spamcast.libsyn.comSparx Systems Enterprise Architect and Ability Engineering EADocX Value Added ResellerOn Tue, Dec 3, 2013 at 10:10 AM, St�phane Ducasse <stephane.ducasse@inria.fr> wrote:On Dec 2, 2013, at 8:46 PM, phil@highoctane.be wrote:And missing a ton of methods.Teh StackInterpreterSimulator is agreeing with this statement.send us the ones that are missing.Please note that we should kill the transcript in the future because this is a ^%$&*((^%$* outdated way to managefeedback: plain string, close the window data is lost, limited in length, no filtering�..bad bad bad does not scale horrible and so frustrating. I will not send you the current transcript I have because I deal with private codebut this does not work.StefOn Mon, Dec 2, 2013 at 8:05 PM, Esteban Lorenzano <estebanlm@gmail.com> wrote:yes, the transcript is ThreadSafeTranscriptOn Mon, Dec 2, 2013 at 8:00 PM, Yuriy Tymchuk <yuriy.tymchuk@me.com> wrote:
Buy the way, is absence of Transcript class - ok?I know how it all works now, but is it ok? :)ukoOn 02 Dec 2013, at 19:58, Esteban Lorenzano <estebanlm@gmail.com> wrote:yes, please :)On Mon, Dec 2, 2013 at 7:56 PM, St�phane Ducasse <stephane.ducasse@inria.fr> wrote:
No I will finish my logger :)
> The return of the toothpick is near. o/
Stef