Peter,
This one is easy to debunk. The idea of mapped diagnostic context was
on the table for a long time. Please see:
Mail dated June 12, 2001:
http://www.mail-archive.com/log4j-user@jakarta.apache.org/msg01446.html
The idea was first exposed in a mail dated 24th of January, 2001.
http://marc.theaimsgroup.com/?l=log4j-dev&m=98051571516266&w=2
These are no opinions, just facts. I hope you will admit that.
The reason why MDCs took so long to implement was conserving JDK 1.1
compatibility. Your release of LogKit with a similar feature,
it just emphasized how important this feature was. I quickly moved to implement
MDCs because I had been thinking about the problem for a long time
and Chris Taylor had just provided a patch for supporting
Thread.getContectClassLoader, a Java2-only feature. I followed up on his idea
to support ThreadLocals in MDC. Is that enough to convince of my good faith?
The same applies to supporting security. We did not make any
SecurityManager checks because of the need to preserve JDK 1.1
compatibility. When we move to an architecture where Logger is just an
interface with different underlying implementations, then we can make
security checks in one implementation and not the other.
Up until now, I don't think anyone in this forum attacked you in any
way Peter. Your claims about security, performance, reliability and so
on are totally unsubstantiated. That is just my opinion.
Regards, Ceki
ps: Charles Darwin in his seminal book "The Origin of Species" observed that
competition between similar species was the fiercest. I think we have an
example of that today.
At 08:23 17.08.2001 +1000, Peter Donald wrote:
>Hi,
>
>I was told to come defend myself here. Way too many messages to read in
>archive and considering nature of the "discussion" I bet I can guess what
>some of the issues were.
>
>Just under a year ago Ceki decided to bring Log4j to Apache. There was some
>legal (non-)issues which took a while to resolve. During this whole time I
>was a very strong advocate of this and sent a bunch of email to Stefano/Ceki
>cheering them on etc.
>
>Ceki asked me to have a look at it and make suggestions. So I spent a few
>hours working over it and sent a bunch of suggestions to Ceki. Most of the
>suggestions were things that LogKit implemented but Log4j didn't. Most of
>these things got implemented. This repeated another time IIRC.
>
>Later Stefano wanted us to use Log4j inside Avalon and drop LogKit. However
>Log4j wasn't (and still isn't) suitable for these environments. It has some
>fairly poor performance characteristerics if you make use of deep
>hierarchies. It is also porous. Once you get access to a Category you get
>access to the whole system. You can masquarade, and mess with logging. Want
>to redirect "auth" messages to /dev/null - easy. And as almost every Log4j
>users uses the default hierarchy ... that is faar worse. You can not trust
>anything the Log4j system logs. I tried to get Ceki to change these things -
>no luck - was stonewalled.
>
>So next I tried to get him interested in rebuilding the whole system - kinda
>like Log4J version 2. He never responded - I assumed it was pointless to try
>to convince him otherwise. Then I tried a last ditch effort to share backend
>(Appenders/LogTargets, Events etc) - but apparently he never received this
>email.
>
>Hence why LogKit continued to exist - however I was still hoping that it or
>Logging JSR would become acceptable. Thanks to Cekis campaign the Logging JSR
>is now good enough unfortunately limited to jdk1.4.
>
>A bit back back Ceki decided to proclaim me a liar, thief and a parasite.
>Apparently those things that LogKit has in common with Log4j were stolen from
>their community (despite being implemented first in LogKit and added into
>Log4j). Ironically even recently, after LogKit made a release with a new
>feature (ContextMaps), withing 24 hours Log4j aquired the equivelent (MDC). I
>assume Ceki would claim they came up with it independently or one of their
>users suggested it or something similar. Curious that I am considered the
>parasite despite originating many of these ideas, huh?
>
>The other thing that may have been trailed out is that LogKit is unstable
>incompatible mess or some trite like that. As far as I am aware LogKit has
>only broken binary compatability 3 times - all of them when it was alpha
>software.
>* once pre-apache
>* once when fede experimented in main tree and changes were later backed out
>* once when change from alpha-beta status
>
>I am sure there is someone willing to call me a liar - but whats the bet they
>arn't willing to provide any proof? ;)
>
>Anyways I just wasted 30 minutes writing this and probably won't respond to
>the trolls that are likely to arise. All I want is a secure fast logging
>toolkit.
>
>--
>Cheers,
>
>Pete
>
>*-----------------------------------------------------*
>* "Faced with the choice between changing one's mind, *
>* and proving that there is no need to do so - almost *
>* everyone gets busy on the proof." *
>* - John Kenneth Galbraith *
>*-----------------------------------------------------*
--
Ceki Gülcü - http://qos.ch
