Re: [boost] [Review] The review of the Boost.Logging library startstoday, November 7th (long)

From: Pavel Vozenilek (pavel_vozenilek_at_[hidden])
Date: 2005-11-13 22:26:00

"John Torjo" wrote:

>> DEFAULT_INDEX /cannot/ be uppercase (clashes with
>> preprocessor).
>
> I understand that it would better be lowercased, but "cannot" is a
> pretty strong word. Care to exemplify?
>
Probability of "DEFAULT_INDEX" being
macro defined in end user code is very high.

_____________________________________________________

>> Dedicated thread used for logging: does it work
>> if both application executable and a DLLs do log into
>> the same file? It is quite tricky to ensure singleton
>> in such circumstances. Jason Hise works on library "Singleton"
>> where he had solved this (after a lot of effort).
>
> I know it's been successfully used on Windows. Would
> Linux/some_other_platform be a problem?
>>

Jason's library allows to have singleton compiled into static library
linked into a DLL whcih is loaded dynamically during session
by EXE. And to do it portably.

He has current code in Sandbox (documentation is being
worked on but he expresses wish to explain what's needed).

Reuse of Singleton library (quite likely to get into Boost
in its second review) would allow to integrate Logger
into wider mesh of singletons

_____________________________________________________

>> 20. Feature: since each log may span to several lines
>>     a modifier could be added to the library:
>>
>>    .add_modifier(log_ending("###"))
>>
>>
>> where every log will be ended with [optional \n if needed]###\n.
>>
>>
>> This would allow easier create tools that manipulate
>> with logs.
>>
>>
>> Similar modifier:
>>    .add_modifier(log_starting("@@@"))
>>
>> where every log starts with @@@ (no newlines)
>> may be considered but this one is not that urgent.
>>
>
> I did not understand what you're saying.
>

I would like to process result logs with my SW.

I am lazy and would like to have clear markers
of log start and end so the SW is easier.

E.g.

<<< 2005/11/24 12:38:43 blah
blah blah
>>>
<<< 2005/11/24 12:38:56 ble ble
>>>

/Pavel