[qos-ch/logback] fd9006: ongoing work on LOGBACK-1509
Branch: refs/heads/master Home: https://github.com/qos-ch/logback Commit: fd9006981ac7ff52fbc79187254b1f3c1bd70bf8 https://github.com/qos-ch/logback/commit/fd9006981ac7ff52fbc79187254b1f3c1bd... Author: Ceki Gulcu <ceki@qos.ch> Date: 2021-08-07 (Sat, 07 Aug 2021) Changed paths: M logback-classic/src/main/java/ch/qos/logback/classic/pattern/NamedConverter.java M logback-classic/src/test/java/ch/qos/logback/classic/pattern/LoggerNameConverterPerfTest.java Log Message: ----------- ongoing work on LOGBACK-1509 Signed-off-by: Ceki Gulcu <ceki@qos.ch>
Hi Its good to see the work on this issue progressing, however there was a solution ready in https://github.com/qos-ch/logback/pull/477 for review, which has been working in a very large closed source systems for over a year now The fix that is not in master seems not to be threadsafe though. LinkedHashMap on which it is based isn't and that why I used CHM Maintaining a threadsafe no blocking LRU is complex, so I would suggest that to implementation in master is at least synchronized I an happy to work on a better lock free solution though, but I am unconvinced that we need to have a limit on the size, and there are simpler approaches to limiting the growth if memory constraints are the concern Regards Mike -----Original Message----- From: logback-dev <logback-dev-bounces@qos.ch> On Behalf Of Ceki Gulcu Sent: 07 August 2021 22:08 To: logback-dev@qos.ch Subject: [logback-dev] [qos-ch/logback] fd9006: ongoing work on LOGBACK-1509 Branch: refs/heads/master Home: https://github.com/qos-ch/logback Commit: fd9006981ac7ff52fbc79187254b1f3c1bd70bf8 https://github.com/qos-ch/logback/commit/fd9006981ac7ff52fbc79187254b1f3c1bd... Author: Ceki Gulcu <ceki@qos.ch> Date: 2021-08-07 (Sat, 07 Aug 2021) Changed paths: M logback-classic/src/main/java/ch/qos/logback/classic/pattern/NamedConverter.java M logback-classic/src/test/java/ch/qos/logback/classic/pattern/LoggerNameConverterPerfTest.java Log Message: ----------- ongoing work on LOGBACK-1509 Signed-off-by: Ceki Gulcu <ceki@qos.ch> _______________________________________________ logback-dev mailing list logback-dev@qos.ch http://mailman.qos.ch/mailman/listinfo/logback-dev
Hi Mike, Thank you for your message as well as PR 477. Please have a look at the code (commit 135bb88cd87e2c9 or later). Note that the viaCache(String) method is synchronized. It is the only place where the cache, a subclass of LinkedHashMap, is modified. It can be cleared via a call to disableCache() but this method may be called indirectly from within viaCache(String). Thus, I think the implementation is quite thread safe. You mentioned a lock free solution as well as solutions limiting memory usage. I am all ears. -- Ceki On 08/08/2021 17:49, mike.skells@talk21.com wrote:
Hi Its good to see the work on this issue progressing, however there was a solution ready in https://github.com/qos-ch/logback/pull/477 for review, which has been working in a very large closed source systems for over a year now The fix that is not in master seems not to be threadsafe though. LinkedHashMap on which it is based isn't and that why I used CHM Maintaining a threadsafe no blocking LRU is complex, so I would suggest that to implementation in master is at least synchronized
I an happy to work on a better lock free solution though, but I am unconvinced that we need to have a limit on the size, and there are simpler approaches to limiting the growth if memory constraints are the concern
Regards Mike
-----Original Message----- From: logback-dev <logback-dev-bounces@qos.ch> On Behalf Of Ceki Gulcu Sent: 07 August 2021 22:08 To: logback-dev@qos.ch Subject: [logback-dev] [qos-ch/logback] fd9006: ongoing work on LOGBACK-1509
Branch: refs/heads/master Home: https://github.com/qos-ch/logback Commit: fd9006981ac7ff52fbc79187254b1f3c1bd70bf8 https://github.com/qos-ch/logback/commit/fd9006981ac7ff52fbc79187254b1f3c1bd... Author: Ceki Gulcu <ceki@qos.ch> Date: 2021-08-07 (Sat, 07 Aug 2021)
Changed paths: M logback-classic/src/main/java/ch/qos/logback/classic/pattern/NamedConverter.java M logback-classic/src/test/java/ch/qos/logback/classic/pattern/LoggerNameConverterPerfTest.java
Log Message: ----------- ongoing work on LOGBACK-1509
Signed-off-by: Ceki Gulcu <ceki@qos.ch>
_______________________________________________ logback-dev mailing list logback-dev@qos.ch http://mailman.qos.ch/mailman/listinfo/logback-dev
_______________________________________________ logback-dev mailing list logback-dev@qos.ch http://mailman.qos.ch/mailman/listinfo/logback-dev
Hi Mike, By the way, there is a new implementation of the TargetLengthBasedClassNameAbbreviator.abbreviate method that is not only simpler, it creates only a single StringBuffer object per call. Comments welcome. -- Ceki On 08/08/2021 18:39, Ceki Gülcü wrote:
Hi Mike,
Thank you for your message as well as PR 477.
Please have a look at the code (commit 135bb88cd87e2c9 or later). Note that the viaCache(String) method is synchronized. It is the only place where the cache, a subclass of LinkedHashMap, is modified. It can be cleared via a call to disableCache() but this method may be called indirectly from within viaCache(String).
Thus, I think the implementation is quite thread safe.
You mentioned a lock free solution as well as solutions limiting memory usage. I am all ears.
-- Ceki
On 08/08/2021 17:49, mike.skells@talk21.com wrote:
Hi Its good to see the work on this issue progressing, however there was a solution ready in https://github.com/qos-ch/logback/pull/477 for review, which has been working in a very large closed source systems for over a year now The fix that is not in master seems not to be threadsafe though. LinkedHashMap on which it is based isn't and that why I used CHM Maintaining a threadsafe no blocking LRU is complex, so I would suggest that to implementation in master is at least synchronized
I an happy to work on a better lock free solution though, but I am unconvinced that we need to have a limit on the size, and there are simpler approaches to limiting the growth if memory constraints are the concern
Regards Mike
-----Original Message----- From: logback-dev <logback-dev-bounces@qos.ch> On Behalf Of Ceki Gulcu Sent: 07 August 2021 22:08 To: logback-dev@qos.ch Subject: [logback-dev] [qos-ch/logback] fd9006: ongoing work on LOGBACK-1509
Branch: refs/heads/master Home: https://github.com/qos-ch/logback Commit: fd9006981ac7ff52fbc79187254b1f3c1bd70bf8
https://github.com/qos-ch/logback/commit/fd9006981ac7ff52fbc79187254b1f3c1bd...
Author: Ceki Gulcu <ceki@qos.ch> Date: 2021-08-07 (Sat, 07 Aug 2021)
Changed paths: M logback-classic/src/main/java/ch/qos/logback/classic/pattern/NamedConverter.java
M logback-classic/src/test/java/ch/qos/logback/classic/pattern/LoggerNameConverterPerfTest.java
Log Message: ----------- ongoing work on LOGBACK-1509
Signed-off-by: Ceki Gulcu <ceki@qos.ch>
participants (3)
-
Ceki Gulcu -
Ceki Gülcü -
mike.skells＠talk21.com