Re: [boost] My pseudo review of Logging library

From: Loïc Joly (loic.actarus.joly_at_[hidden])
Date: 2008-02-11 16:13:16

Gennadiy Rozental a écrit :

I have not been able to devote enough time to make a full length review. 
So this is more of a feeling than of a review. Fell free to ignome my 
vote accordingly.
> * What is your evaluation of the design?
>   
It seems to be a little bit too raw to be usable by the unwary user.
Some eviewers have complained about the use of macro. Loggoing is IMO 
one of the few places where macro usage is a good thing. There are some 
places where logging won't get acceptance if it cannot be demonstrated 
the it can be zero cost when de-activated (even if all odds are that it 
will finally not be de-activated). It is a psychological problem as well 
as a technical problem, but it is important.
> * What is your evaluation of the implementation?
>   
I did not look at the implementation.
> * What is your evaluation of the documentation?
>   
The learning curve is too steep. Is it the fault of the documentation or 
of the library ? I do not know.
> * What is your evaluation of the potential usefulness of the library?
>   
Huge.
> * Did you try to use the library? With what compiler? Did you have any
> problems?
>   
Not that version. I use previous version successfully with visual C++, 
but this version sees to be so much appart that I do not know if this is 
relevant at all.
> * How much effort did you put into your evaluation? A glance? A quick
> reading?
> In-depth study?
>   
Only read the doc, about 1h spent
> * Are you knowledgeable about the problem domain?
>   
Only as well as any programmer should be.
> And finally, every review should answer this question:
>
> * Do you think the library should be accepted as a Boost library
I would like to agree with other reviewers: Logging is so vast a project 
that anyone want it own special options that fit its particular needs. I 
think the review of the previous version of the library suffered the 
same problem. This has led John Torjo to write the swiss army knife of 
logging, and still, some people miss miss some options... My opinion is 
that a logging library is so much needed that I do not really care if it 
is not perfect, as long as it is usable & extendable.

What I miss most is simplicity. I would really appreciate is a basig 
logging configuration was provided for a quickstart. By basic, I mean 
something that allow code such a this :
- Log onto a file
- Set the log level (at compile time or at run time)
- Log using  iostream syntax
- Initialization should be a one-liner in user code.
It can allow more, but the one-liner stuff is important.

I don't knows if there if a fast track second chance review process for 
a library. If there is one, I think it would be better to wait a few 
weeks and have another review. Otherwise, I vote yes, knowing this 
library will probably evolve anyway after the many remarks done during 
the review.

-- 
Loïc Joly