Re: possible to filter the output to commits@ list????

2010-10-25 Thread Grant Ingersoll
Hoss, you and I should see if we can wrangle some of the SVN guys at ApacheCon 
and see if we can get this stuff straightened out.

On Oct 24, 2010, at 8:33 PM, Yonik Seeley wrote:

 Another downside to the current situation - svn log shows all these
 updates to the mergeprops, so we're losing the real updates in the
 noise.
 
 Maybe we should try to merge only from trunk to branch_3x, so at least
 trunk says relatively clean.  Making sense of branch_3x is getting
 very difficult.
 
 Example:
 yo...@wolverine /cygdrive/c/code/lusolr_3x
 $ svn log lucene/src/test/org/apache/lucene/search/spans/TestPayloadSpans.java
 
 r1026921 | yonik | 2010-10-24 20:23:17 -0400 (Sun, 24 Oct 2010) | 1 line
 
 tests: fix resource leaks (merge from trunk)
 
 r1026885 | yonik | 2010-10-24 16:56:54 -0400 (Sun, 24 Oct 2010) | 1 line
 
 SOLR-2192: sync runners in StreamingUpdateSolrServer.blockUntilFinished
 
 r1026873 | yonik | 2010-10-24 15:52:34 -0400 (Sun, 24 Oct 2010) | 1 line
 
 SOLR-2160: tests - use classloader from core resource loader
 
 r1026869 | yonik | 2010-10-24 15:24:39 -0400 (Sun, 24 Oct 2010) | 1 line
 
 SOLR-2162: tests should use multi-threaded connection manager
 
 r1026739 | koji | 2010-10-24 00:42:39 -0400 (Sun, 24 Oct 2010) | 1 line
 
 fix indent in DIH sample
 
 r1026593 | mikemccand | 2010-10-23 06:23:02 -0400 (Sat, 23 Oct 2010) | 1 line
 
 LUCENE-2618: allow optimize to complete during IW.close, take 2
 
 r1026432 | mikemccand | 2010-10-22 13:59:16 -0400 (Fri, 22 Oct 2010) | 1 line
 
 LUCENE-2618: revert
 
 r1026337 | mikemccand | 2010-10-22 10:17:10 -0400 (Fri, 22 Oct 2010) | 1 line
 
 LUCENE-2618: make sure optimize merge complete even if a close is pending
 
 r1026170 | uschindler | 2010-10-21 18:39:44 -0400 (Thu, 21 Oct 2010) | 1 line
 
 merge: remove warning
 
 r1026053 | yonik | 2010-10-21 12:16:22 -0400 (Thu, 21 Oct 2010) | 1 line
 
 SOLR-2180: close request in finally block
 
 r1026035 | yonik | 2010-10-21 11:49:45 -0400 (Thu, 21 Oct 2010) | 1 line
 
 tests: fix resource leaks, convert to Junit4 (merged from trunk)
 
 r1026024 | yonik | 2010-10-21 11:09:32 -0400 (Thu, 21 Oct 2010) | 1 line
 
 axe extra distrib test (merge trunk 1022805)
 
 r1026000 | gsingers | 2010-10-21 09:48:34 -0400 (Thu, 21 Oct 2010) | 1 line
 
 SOLR-2010, including Yonik's fix, SOLR-2181 -- hope I did this merge correctly
 
 r1025931 | mikemccand | 2010-10-21 06:29:01 -0400 (Thu, 21 Oct 2010) | 1 line
 
 fix MTQ.CutOffTermCollector to check limits after adding term, not before
 
 r1025606 | rmuir | 2010-10-20 10:51:12 -0400 (Wed, 20 Oct 2010) | 1 line
 
 LUCENE-1938: Precedence query parser using the contrib/queryparser framework
 
 r1024481 | yonik | 2010-10-19 21:27:57 -0400 (Tue, 19 Oct 2010) | 1 line
 
 SOLR-2197: wait for search executor to close before closing main
 server (merge trunk 1024476)
 
 r1024409 | uschindler | 2010-10-19 16:57:52 -0400 (Tue, 19 Oct 2010) | 1 line
 
 LUCENE-2556: Improve memory usage after cloning TermAttribute
 
 r1024405 | rmuir | 2010-10-19 16:46:23 -0400 (Tue, 19 Oct 2010) | 1 line
 
 LUCENE-1937: add methods to manipulate QueryNodeProcessorPipeline elements
 
 r1024258 | rmuir | 2010-10-19 11:04:11 -0400 (Tue, 19 Oct 2010) | 1 line
 
 SOLR-1794: Dataimport of CLOB fields fails when getCharacterStream()
 is defined in a superclass
 
 r1024223 | rmuir | 2010-10-19 08:59:48 -0400 (Tue, 19 Oct 2010) | 1 line
 
 LUCENE-2652: remove obselete runners, they are inconsistent with
 @beforeClass and obselete in the build today
 

Re: possible to filter the output to commits@ list????

2010-10-24 Thread Yonik Seeley
Another downside to the current situation - svn log shows all these
updates to the mergeprops, so we're losing the real updates in the
noise.

Maybe we should try to merge only from trunk to branch_3x, so at least
trunk says relatively clean.  Making sense of branch_3x is getting
very difficult.

Example:
yo...@wolverine /cygdrive/c/code/lusolr_3x
$ svn log lucene/src/test/org/apache/lucene/search/spans/TestPayloadSpans.java

r1026921 | yonik | 2010-10-24 20:23:17 -0400 (Sun, 24 Oct 2010) | 1 line

tests: fix resource leaks (merge from trunk)

r1026885 | yonik | 2010-10-24 16:56:54 -0400 (Sun, 24 Oct 2010) | 1 line

SOLR-2192: sync runners in StreamingUpdateSolrServer.blockUntilFinished

r1026873 | yonik | 2010-10-24 15:52:34 -0400 (Sun, 24 Oct 2010) | 1 line

SOLR-2160: tests - use classloader from core resource loader

r1026869 | yonik | 2010-10-24 15:24:39 -0400 (Sun, 24 Oct 2010) | 1 line

SOLR-2162: tests should use multi-threaded connection manager

r1026739 | koji | 2010-10-24 00:42:39 -0400 (Sun, 24 Oct 2010) | 1 line

fix indent in DIH sample

r1026593 | mikemccand | 2010-10-23 06:23:02 -0400 (Sat, 23 Oct 2010) | 1 line

LUCENE-2618: allow optimize to complete during IW.close, take 2

r1026432 | mikemccand | 2010-10-22 13:59:16 -0400 (Fri, 22 Oct 2010) | 1 line

LUCENE-2618: revert

r1026337 | mikemccand | 2010-10-22 10:17:10 -0400 (Fri, 22 Oct 2010) | 1 line

LUCENE-2618: make sure optimize merge complete even if a close is pending

r1026170 | uschindler | 2010-10-21 18:39:44 -0400 (Thu, 21 Oct 2010) | 1 line

merge: remove warning

r1026053 | yonik | 2010-10-21 12:16:22 -0400 (Thu, 21 Oct 2010) | 1 line

SOLR-2180: close request in finally block

r1026035 | yonik | 2010-10-21 11:49:45 -0400 (Thu, 21 Oct 2010) | 1 line

tests: fix resource leaks, convert to Junit4 (merged from trunk)

r1026024 | yonik | 2010-10-21 11:09:32 -0400 (Thu, 21 Oct 2010) | 1 line

axe extra distrib test (merge trunk 1022805)

r1026000 | gsingers | 2010-10-21 09:48:34 -0400 (Thu, 21 Oct 2010) | 1 line

SOLR-2010, including Yonik's fix, SOLR-2181 -- hope I did this merge correctly

r1025931 | mikemccand | 2010-10-21 06:29:01 -0400 (Thu, 21 Oct 2010) | 1 line

fix MTQ.CutOffTermCollector to check limits after adding term, not before

r1025606 | rmuir | 2010-10-20 10:51:12 -0400 (Wed, 20 Oct 2010) | 1 line

LUCENE-1938: Precedence query parser using the contrib/queryparser framework

r1024481 | yonik | 2010-10-19 21:27:57 -0400 (Tue, 19 Oct 2010) | 1 line

SOLR-2197: wait for search executor to close before closing main
server (merge trunk 1024476)

r1024409 | uschindler | 2010-10-19 16:57:52 -0400 (Tue, 19 Oct 2010) | 1 line

LUCENE-2556: Improve memory usage after cloning TermAttribute

r1024405 | rmuir | 2010-10-19 16:46:23 -0400 (Tue, 19 Oct 2010) | 1 line

LUCENE-1937: add methods to manipulate QueryNodeProcessorPipeline elements

r1024258 | rmuir | 2010-10-19 11:04:11 -0400 (Tue, 19 Oct 2010) | 1 line

SOLR-1794: Dataimport of CLOB fields fails when getCharacterStream()
is defined in a superclass

r1024223 | rmuir | 2010-10-19 08:59:48 -0400 (Tue, 19 Oct 2010) | 1 line

LUCENE-2652: remove obselete runners, they are inconsistent with
@beforeClass and obselete in the build today

r1024197 | mikemccand | 2010-10-19 06:27:19 -0400 (Tue, 19 Oct 2010) | 1 line

add some missing super.setParams so alg.toString shows the params

r1023877 | uschindler | 2010-10-18 

Re: possible to filter the output to commits@ list????

2010-10-20 Thread Grant Ingersoll

On Oct 19, 2010, at 4:14 PM, Robert Muir wrote:

 On Tue, Oct 19, 2010 at 3:51 PM, Michael McCandless
 luc...@mikemccandless.com wrote:
 
 Hmm but how does this explain how the number of mergeprops has grown
 so much recently?  I think recent merges have generally been toplevel?
 
 
 i introduced a lot of them by combining trunk's analyzers into 
 modules/analysis,
 because things were reorganized, there are many, but still there are
 more than necessary and this could be cleaned up.

Couldn't 3.x move contrib/analysis to modules?  We could just move it to 
contrib as part of the release packaging  if the goal is to be consistent, but 
otherwise who cares where it is located?

Would that reduce some of this?

 
 still, if we are going to refactor lucene+lucene contrib+solr in other
 ways (i hope we will!) I would imagine we would see more of this in
 the future.
 
 -
 To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
 For additional commands, e-mail: dev-h...@lucene.apache.org
 

--
Grant Ingersoll
http://www.lucidimagination.com/

Search the Lucene ecosystem docs using Solr/Lucene:
http://www.lucidimagination.com/search


-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



RE: possible to filter the output to commits@ list????

2010-10-20 Thread Uwe Schindler
No. Once the files have a property they have it irreversible. This is why
single-file merges are a no-go.

-
Uwe Schindler
H.-H.-Meier-Allee 63, D-28213 Bremen
http://www.thetaphi.de
eMail: u...@thetaphi.de


 -Original Message-
 From: Grant Ingersoll [mailto:gsing...@apache.org]
 Sent: Wednesday, October 20, 2010 1:01 PM
 To: dev@lucene.apache.org
 Subject: Re: possible to filter the output to commits@ list
 
 
 On Oct 19, 2010, at 4:14 PM, Robert Muir wrote:
 
  On Tue, Oct 19, 2010 at 3:51 PM, Michael McCandless
  luc...@mikemccandless.com wrote:
 
  Hmm but how does this explain how the number of mergeprops has grown
  so much recently?  I think recent merges have generally been toplevel?
 
 
  i introduced a lot of them by combining trunk's analyzers into
  modules/analysis, because things were reorganized, there are many, but
  still there are more than necessary and this could be cleaned up.
 
 Couldn't 3.x move contrib/analysis to modules?  We could just move it to
 contrib as part of the release packaging  if the goal is to be consistent,
but
 otherwise who cares where it is located?
 
 Would that reduce some of this?
 
 
  still, if we are going to refactor lucene+lucene contrib+solr in other
  ways (i hope we will!) I would imagine we would see more of this in
  the future.
 
  -
  To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For
  additional commands, e-mail: dev-h...@lucene.apache.org
 
 
 --
 Grant Ingersoll
 http://www.lucidimagination.com/
 
 Search the Lucene ecosystem docs using Solr/Lucene:
 http://www.lucidimagination.com/search
 
 
 -
 To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional
 commands, e-mail: dev-h...@lucene.apache.org



-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-20 Thread Robert Muir
On Wed, Oct 20, 2010 at 7:01 AM, Grant Ingersoll gsing...@apache.org wrote:

 Couldn't 3.x move contrib/analysis to modules?  We could just move it to 
 contrib as part of the release packaging  if the goal is to be consistent, 
 but otherwise who cares where it is located?

 Would that reduce some of this?


I don't think it would help so much, because the analyzers in modules
did not come just from contrib/analyzers... but from lucene's core
(individual files cherrypicked, things like TokenStream still stay),
contrib/icu, and from Solr (the factories stay at the moment though).

thats just what i mean though, if we will make lucene more modular i
think this is just a very typical case of consolidating things from
different places in lucene/solr and modularizing it, so thats why I
think a lot of it is unavoidable.

Uwe says single-file merges are a no-go, but (hypothetically) if we
moved just the concrete Queries from o.a.l.search (but not everything
else!), and the Queries in contrib/queries into a modules/queries,
then you would end up with the same situation... if you fix a bug in
modules/queries/TermQuery you have to merge the individual file in
3.x's src/java because the entire o.a.l.search did not move, only a
few cherrypicked files (the queries).

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-19 Thread Michael McCandless
On Mon, Oct 18, 2010 at 3:03 PM, Grant Ingersoll gsing...@apache.org wrote:

  It was just a bit of a shocker for me the first time I did it and I see like 
 30 files changed when I only changed one file.

Me too.  In fact I think it's ridiculous -- violates principle of
least surprise.  I shouldn't have to see such details of the source
control system's impl

Furthermore, I think it may eventually turn into a serious perf issue,
since it seems to be an O(N^2) growth.  We are at 7 emails today,
which was only 3 emails not long ago.  Where will be be a few months
from now?  (Though I guess it is bounded by the total number of source
files we have in 3.x...).

Maybe svn is trying to tell us to release 4.0, heh ;)

Mike

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-19 Thread Upayavira
FWIW, the commit notices are just an SVN post-commit hook that uses the
svn-mailer tool [http://opensource.perlig.de/svnmailer/]. I believe
Grant has commit rights to that file - it is in the infra SVN right next
to the old asf-auth file.

Right now I have no idea whether svn-mailer can support the kind of
filtering you are talking about, but there's no harm looking!

Upayavira

On Tue, 19 Oct 2010 06:45 -0400, Michael McCandless
luc...@mikemccandless.com wrote:
 On Mon, Oct 18, 2010 at 3:03 PM, Grant Ingersoll gsing...@apache.org
 wrote:
 
   It was just a bit of a shocker for me the first time I did it and I see 
  like 30 files changed when I only changed one file.
 
 Me too.  In fact I think it's ridiculous -- violates principle of
 least surprise.  I shouldn't have to see such details of the source
 control system's impl
 
 Furthermore, I think it may eventually turn into a serious perf issue,
 since it seems to be an O(N^2) growth.  We are at 7 emails today,
 which was only 3 emails not long ago.  Where will be be a few months
 from now?  (Though I guess it is bounded by the total number of source
 files we have in 3.x...).
 
 Maybe svn is trying to tell us to release 4.0, heh ;)
 
 Mike
 
 -
 To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
 For additional commands, e-mail: dev-h...@lucene.apache.org
 

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-19 Thread Grant Ingersoll

On Oct 19, 2010, at 7:06 AM, Upayavira wrote:

 FWIW, the commit notices are just an SVN post-commit hook that uses the
 svn-mailer tool [http://opensource.perlig.de/svnmailer/]. I believe
 Grant has commit rights to that file - it is in the infra SVN right next
 to the old asf-auth file.

Yes, I do, but it isn't clear if this is something that could be handled 
through simple routing.  I suppose one thing we could do is filter it through a 
dummy GMail account somehow.  Send all commits to the GMail group and then have 
it send a digest email a few times a day to comm...@l.a.o.  It's ugly, but it 
might work.

FWIW, I get the sense that a lot of other projects deal with merges.  What do 
they do?

 
 Right now I have no idea whether svn-mailer can support the kind of
 filtering you are talking about, but there's no harm looking!
 
 Upayavira
 
 On Tue, 19 Oct 2010 06:45 -0400, Michael McCandless
 luc...@mikemccandless.com wrote:
 On Mon, Oct 18, 2010 at 3:03 PM, Grant Ingersoll gsing...@apache.org
 wrote:
 
  It was just a bit of a shocker for me the first time I did it and I see 
 like 30 files changed when I only changed one file.
 
 Me too.  In fact I think it's ridiculous -- violates principle of
 least surprise.  I shouldn't have to see such details of the source
 control system's impl
 
 Furthermore, I think it may eventually turn into a serious perf issue,
 since it seems to be an O(N^2) growth.  We are at 7 emails today,
 which was only 3 emails not long ago.  Where will be be a few months
 from now?  (Though I guess it is bounded by the total number of source
 files we have in 3.x...).
 
 Maybe svn is trying to tell us to release 4.0, heh ;)
 
 Mike
 
 -
 To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
 For additional commands, e-mail: dev-h...@lucene.apache.org
 
 
 -
 To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
 For additional commands, e-mail: dev-h...@lucene.apache.org
 

--
Grant Ingersoll
http://www.lucidimagination.com/

Search the Lucene ecosystem docs using Solr/Lucene:
http://www.lucidimagination.com/search


-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-19 Thread Chris Hostetter

: FWIW, I get the sense that a lot of other projects deal with merges.  What do 
they do?

I suspect they do merges properly and avoid this problem entirely.

Bottom Line: if *all* merges happen at the top level, then this problem 
won't exist -- mergeinfo props get added to individual files only when 
individual file merges take place, or with merges from mixed working 
revisions.  Once a file gets those merge props, all future merges 
in that tree interact with those individual file props even if the merge 
has nothing to do with those files...

http://svnbook.red-bean.com/en/1.5/svn.branchmerge.advanced.html#svn.branchmerge.advanced.finalword

 For long-lived release branches (as described in the section called 
 “Common Branching Patterns”), perform merges only on the root of 
 the branch, not on subdirectories.

-Hoss

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

Re: possible to filter the output to commits@ list????

2010-10-19 Thread Grant Ingersoll

On Oct 19, 2010, at 3:04 PM, Chris Hostetter wrote:

 
 : FWIW, I get the sense that a lot of other projects deal with merges.  What 
 do they do?
 
 I suspect they do merges properly and avoid this problem entirely.
 
 Bottom Line: if *all* merges happen at the top level, then this problem 
 won't exist -- mergeinfo props get added to individual files only when 
 individual file merges take place, or with merges from mixed working 
 revisions.  Once a file gets those merge props, all future merges 
 in that tree interact with those individual file props even if the merge 
 has nothing to do with those files...
 
 http://svnbook.red-bean.com/en/1.5/svn.branchmerge.advanced.html#svn.branchmerge.advanced.finalword

I will go take a look at it, but as far as I could tell, I merged at the top 
level, but maybe IntelliJ does individual file merges and I wasn't aware of it 
(I believe I said merge trunk to branch_3x and then picked which revisions I 
wanted to apply).  Perhaps we should put together some examples in Lucene's 
context that would give Lucene developers a deeper understanding.

-Grant
-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



RE: possible to filter the output to commits@ list????

2010-10-19 Thread Uwe Schindler
Hi Grant,

The individual merges are there since long time. And the additional ones
came in later. As I said before (answer to first mail on this thread): The
problem is that once a file gets a separate merge property not inherited
from its parent folder, this merge prop must be updated on every later
merge. So whenever you merge a file separately you create one more problem
:-)

Some merges in trunk cannot be done completely top-level, but if they are
always done in the same way, the number of files/subdirs with separate
mergeprops is limited. One problem is the new modules folder as the analysis
folder in it must be merged to contrib. Also the moved tokenstreams and
analyzers. But all other touched files are results of one-file merges.

Uwe

-
Uwe Schindler
H.-H.-Meier-Allee 63, D-28213 Bremen
http://www.thetaphi.de
eMail: u...@thetaphi.de


 -Original Message-
 From: Grant Ingersoll [mailto:gsing...@apache.org]
 Sent: Tuesday, October 19, 2010 9:30 PM
 To: dev@lucene.apache.org
 Subject: Re: possible to filter the output to commits@ list
 
 
 On Oct 19, 2010, at 3:04 PM, Chris Hostetter wrote:
 
 
  : FWIW, I get the sense that a lot of other projects deal with merges.
What
 do they do?
 
  I suspect they do merges properly and avoid this problem entirely.
 
  Bottom Line: if *all* merges happen at the top level, then this
  problem won't exist -- mergeinfo props get added to individual files
  only when individual file merges take place, or with merges from mixed
  working revisions.  Once a file gets those merge props, all future
  merges in that tree interact with those individual file props even if
  the merge has nothing to do with those files...
 
  http://svnbook.red-bean.com/en/1.5/svn.branchmerge.advanced.html#svn.b
  ranchmerge.advanced.finalword
 
 I will go take a look at it, but as far as I could tell, I merged at the
top level, but
 maybe IntelliJ does individual file merges and I wasn't aware of it (I
believe I
 said merge trunk to branch_3x and then picked which revisions I wanted to
 apply).  Perhaps we should put together some examples in Lucene's context
 that would give Lucene developers a deeper understanding.
 
 -Grant
 -
 To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional
 commands, e-mail: dev-h...@lucene.apache.org



-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-19 Thread Michael McCandless
On Tue, Oct 19, 2010 at 3:29 PM, Grant Ingersoll gsing...@apache.org wrote:

 I will go take a look at it, but as far as I could tell, I merged at the top 
 level, but maybe IntelliJ does individual file merges and I wasn't aware of 
 it (I believe I said merge trunk to branch_3x and then picked which revisions 
 I wanted to apply).  Perhaps we should put together some examples in Lucene's 
 context that would give Lucene developers a deeper understanding.

It's not just you... it's any of us in the past who merged singleton
files (I'm pretty sure I did so before I understood this must merge
from root mantra).

Hmm but how does this explain how the number of mergeprops has grown
so much recently?  I think recent merges have generally been toplevel?

Mike

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-19 Thread Robert Muir
On Tue, Oct 19, 2010 at 3:51 PM, Michael McCandless
luc...@mikemccandless.com wrote:

 Hmm but how does this explain how the number of mergeprops has grown
 so much recently?  I think recent merges have generally been toplevel?


i introduced a lot of them by combining trunk's analyzers into modules/analysis,
because things were reorganized, there are many, but still there are
more than necessary and this could be cleaned up.

still, if we are going to refactor lucene+lucene contrib+solr in other
ways (i hope we will!) I would imagine we would see more of this in
the future.

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



RE: possible to filter the output to commits@ list????

2010-10-19 Thread Uwe Schindler
Hi Upayavira,

Thanks for the hint. Indeed with changing the config file (which allows
special configs for specific subtrees of the svn, so we can do it only for
Lucene), we can do it very easy:

http://opensource.perlig.de/svnmailer/doc-1.0/#groups-generate-diffs

The generate_diffs option defines which actions diffs are generated for. It
takes a space or tab separated list of one or more of the following tokens:
add, modify, copy, delete, propchange and none.

If the add token is given and a new file is added to the repository, the
svnmailer generates a diff between an empty file and the newly added one. If
the modify token is given and the content of an already existing file is
changed, a diff between the old revision and the new revision of that file
is generated. The copy token only worries about files, that are copied and
modified during one commit. The delete token generates a diff between the
previous revision of the file and an empty file, if a file was deleted.

If the propchange token is given, the svnmailer also takes care of changes
in versioned properties. Whether it should actually generate diffs for the
property change action depends on the other tokens of the generate_diffs
list. The same rules as for files apply, except that the svnmailer never
generates property diffs for deleted files


If we change that config option and remove propchange, then the diffs would
not contain propchanges anymore. It would it only list as modified files,
but with that we can live.

Grant: Can you send me a copy of the current config file of that tool? I
could create a patch! (I am allowed to see it).

Uwe

-
Uwe Schindler
H.-H.-Meier-Allee 63, D-28213 Bremen
http://www.thetaphi.de
eMail: u...@thetaphi.de


 -Original Message-
 From: Upayavira [mailto:u...@odoko.co.uk]
 Sent: Tuesday, October 19, 2010 1:07 PM
 To: dev@lucene.apache.org
 Subject: Re: possible to filter the output to commits@ list
 
 FWIW, the commit notices are just an SVN post-commit hook that uses the
svn-
 mailer tool [http://opensource.perlig.de/svnmailer/]. I believe Grant has
 commit rights to that file - it is in the infra SVN right next to the old
asf-auth
 file.
 
 Right now I have no idea whether svn-mailer can support the kind of
filtering
 you are talking about, but there's no harm looking!
 
 Upayavira
 
 On Tue, 19 Oct 2010 06:45 -0400, Michael McCandless
 luc...@mikemccandless.com wrote:
  On Mon, Oct 18, 2010 at 3:03 PM, Grant Ingersoll gsing...@apache.org
  wrote:
 
    It was just a bit of a shocker for me the first time I did it and I
see like 30
 files changed when I only changed one file.
 
  Me too.  In fact I think it's ridiculous -- violates principle of
  least surprise.  I shouldn't have to see such details of the source
  control system's impl
 
  Furthermore, I think it may eventually turn into a serious perf issue,
  since it seems to be an O(N^2) growth.  We are at 7 emails today,
  which was only 3 emails not long ago.  Where will be be a few months
  from now?  (Though I guess it is bounded by the total number of source
  files we have in 3.x...).
 
  Maybe svn is trying to tell us to release 4.0, heh ;)
 
  Mike
 
  -
  To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For
  additional commands, e-mail: dev-h...@lucene.apache.org
 
 
 -
 To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional
 commands, e-mail: dev-h...@lucene.apache.org



-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



RE: possible to filter the output to commits@ list????

2010-10-19 Thread Uwe Schindler
 (I am allowed to see it).

I mean: (if I am allowed to see it.)

-
Uwe Schindler
H.-H.-Meier-Allee 63, D-28213 Bremen
http://www.thetaphi.de
eMail: u...@thetaphi.de


 -Original Message-
 From: Uwe Schindler [mailto:u...@thetaphi.de]
 Sent: Tuesday, October 19, 2010 11:21 PM
 To: dev@lucene.apache.org
 Subject: RE: possible to filter the output to commits@ list
 
 Hi Upayavira,
 
 Thanks for the hint. Indeed with changing the config file (which allows
special
 configs for specific subtrees of the svn, so we can do it only for
Lucene), we can
 do it very easy:
 
 http://opensource.perlig.de/svnmailer/doc-1.0/#groups-generate-diffs
 
 The generate_diffs option defines which actions diffs are generated for.
It
 takes a space or tab separated list of one or more of the following
tokens:
 add, modify, copy, delete, propchange and none.
 
 If the add token is given and a new file is added to the repository, the
svnmailer
 generates a diff between an empty file and the newly added one. If the
modify
 token is given and the content of an already existing file is changed, a
diff
 between the old revision and the new revision of that file is generated.
The
 copy token only worries about files, that are copied and modified during
one
 commit. The delete token generates a diff between the previous revision of
the
 file and an empty file, if a file was deleted.
 
 If the propchange token is given, the svnmailer also takes care of changes
in
 versioned properties. Whether it should actually generate diffs for the
property
 change action depends on the other tokens of the generate_diffs list. The
same
 rules as for files apply, except that the svnmailer never generates
property diffs
 for deleted files
 
 
 If we change that config option and remove propchange, then the diffs
would
 not contain propchanges anymore. It would it only list as modified files,
but
 with that we can live.
 
 Grant: Can you send me a copy of the current config file of that tool? I
could
 create a patch! (I am allowed to see it).
 
 Uwe
 
 -
 Uwe Schindler
 H.-H.-Meier-Allee 63, D-28213 Bremen
 http://www.thetaphi.de
 eMail: u...@thetaphi.de
 
 
  -Original Message-
  From: Upayavira [mailto:u...@odoko.co.uk]
  Sent: Tuesday, October 19, 2010 1:07 PM
  To: dev@lucene.apache.org
  Subject: Re: possible to filter the output to commits@ list
 
  FWIW, the commit notices are just an SVN post-commit hook that uses
  the
 svn-
  mailer tool [http://opensource.perlig.de/svnmailer/]. I believe Grant
  has commit rights to that file - it is in the infra SVN right next to
  the old
 asf-auth
  file.
 
  Right now I have no idea whether svn-mailer can support the kind of
 filtering
  you are talking about, but there's no harm looking!
 
  Upayavira
 
  On Tue, 19 Oct 2010 06:45 -0400, Michael McCandless
  luc...@mikemccandless.com wrote:
   On Mon, Oct 18, 2010 at 3:03 PM, Grant Ingersoll
   gsing...@apache.org
   wrote:
  
 It was just a bit of a shocker for me the first time I did it and
I
 see like 30
  files changed when I only changed one file.
  
   Me too.  In fact I think it's ridiculous -- violates principle of
   least surprise.  I shouldn't have to see such details of the source
   control system's impl
  
   Furthermore, I think it may eventually turn into a serious perf
   issue, since it seems to be an O(N^2) growth.  We are at 7 emails
   today, which was only 3 emails not long ago.  Where will be be a few
   months from now?  (Though I guess it is bounded by the total number
   of source files we have in 3.x...).
  
   Maybe svn is trying to tell us to release 4.0, heh ;)
  
   Mike
  
   
   - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For
   additional commands, e-mail: dev-h...@lucene.apache.org
  
 
  -
  To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For
  additional commands, e-mail: dev-h...@lucene.apache.org
 
 
 
 -
 To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional
 commands, e-mail: dev-h...@lucene.apache.org



-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-18 Thread Grant Ingersoll

On Oct 17, 2010, at 10:58 AM, Robert Muir wrote:

 Lets say i change a single line of code, and merge it back to the 3x_branch.
 
 Currently we get 6 or 7 emails of mergeproperty changes to the
 commits@ list... this is making it difficult or impossible for
 backports to be reviewed at all... I think this is terrible.

I find it onerous that one need do a merge for this kind of thing period.  Why 
not just apply the patch a second time?  Sure, something is lost in SVN, but 
it's covered elsewhere.  Of course, the flip side is that by not doing it, it 
becomes all that much harder to merge in the future.


-Grant
-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-18 Thread Grant Ingersoll
What commands do you want me to run?

On Oct 17, 2010, at 11:27 AM, Uwe Schindler wrote:

 Hi Daniel,
 
 I already proposed something similar to our developers list and also
 explained, why it is a bad idea to merge single files (see attached mail).
 
 Still our/my opinion: It may be better, to simply provide optionally plain
 diffs without metadata for commit mails (on request of a project, e.g. Grant
 Ingersoll could configure that for the Lucene project in the SVN config
 file).
 
 Uwe
 
 -
 Uwe Schindler
 uschind...@apache.org 
 Apache Lucene PMC Member / Committer
 Bremen, Germany
 http://lucene.apache.org/
 
 
 -Original Message-
 From: Daniel Shahaf [mailto:d...@daniel.shahaf.name]
 Sent: Sunday, October 17, 2010 5:17 PM
 To: Uwe Schindler
 Cc: dev@lucene.apache.org; 'Apache Infrastructure'
 Subject: Re: possible to filter the output to commits@ list
 
 Or you could try to see if some of the existing mergeinfo can be removed,
 and
 try to have less subtree mergeinfo in the first place; these are common
 topics
 on us...@subversion.
 
 Uwe Schindler wrote on Sun, Oct 17, 2010 at 17:06:44 +0200:
 Hi all,
 
 There are configuration options in SVN to let the commit mails pipe
 through a
 filter. Other svn hosting providers like Sourceforge provide some filters
 to be
 applied. Ideally, you would use the patchutils package (it's called like
 this in
 Debian/Ubuntu) and use svn diff | filterdiff --clean and pipe that into
 an
 eMail. Maybe we should open an issue at infra?
 
 Uwe
 
 -
 Uwe Schindler
 uschind...@apache.org
 Apache Lucene PMC Member / Committer
 Bremen, Germany
 http://lucene.apache.org/
 
 -Original Message-
 From: Robert Muir [mailto:rcm...@gmail.com]
 Sent: Sunday, October 17, 2010 4:58 PM
 To: dev@lucene.apache.org
 Subject: possible to filter the output to commits@ list
 
 Lets say i change a single line of code, and merge it back to the
 3x_branch.
 
 Currently we get 6 or 7 emails of mergeproperty changes to the
 commits@ list... this is making it difficult or impossible for
 backports to be reviewed at all... I think this is terrible.
 
 How can we get this fixed so that these mergeprop-changes only are
 filtered in the emails? Someone could always click the link to
 viewvc or whatever if they are interested in this...
 
 
 - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For
 additional commands, e-mail: dev-h...@lucene.apache.org
 
 
 Mail Attachment.eml

--
Grant Ingersoll
http://www.lucidimagination.com


-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-18 Thread Robert Muir
On Mon, Oct 18, 2010 at 12:47 PM, Grant Ingersoll gsing...@apache.org wrote:

 I find it onerous that one need do a merge for this kind of thing period.  
 Why not just apply the patch a second time?  Sure, something is lost in SVN, 
 but it's covered elsewhere.  Of course, the flip side is that by not doing 
 it, it becomes all that much harder to merge in the future.


why use version control at all?

by merging, the practical benefit to me is it tracks what has and
hasn't been merged (my ide has a nice interface that lets me quickly
see eligible revs by my username, etc)

but if you just patch and commit twice, it looks to svn from the
mergeinfo that you never merged that change back to branch_3x... if
you are going to do this, then at least mark the revision as merged so
it doesnt mislead people who are using version control.

(separately merging makes life easier to me for anything non-trivial,
a patch wouldn't apply cleanly anyway to both our branches, i'd rather
just resolve the conflicting parts)

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-18 Thread Chris Hostetter

: I find it onerous that one need do a merge for this kind of thing 
: period.  Why not just apply the patch a second time?  Sure, something is 
: lost in SVN, but it's covered elsewhere.  Of course, the flip side is 
: that by not doing it, it becomes all that much harder to merge in the 
: future.

Exactly -- the amount of work required to apply a one line patch to two 
distinct branches is about the same as the amount of work to apply that 
patch to trunk and then svn merge it to the branch -- but the impact on 
other people who have much more complicated merges is huge.  when you use 
svn merge, you make all of those future (big) merges much simpler to deal 
with -- but every time you apply a patch directly to both branches and 
commit individually, you introduce a conflict that has to be resolved 
manually by every person who ever tries an svn merge on that file in the 
future.


-Hoss

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



RE: possible to filter the output to commits@ list????

2010-10-18 Thread Uwe Schindler
Hi Grant,

Not yet decided/investigated, this was more a question to infra, if it would
be possible to set this up. :-) I have to check a config from an SVN server
that supports merging. Where are the Apach config files for SVN? Can I view
them?

Else, I agree with Robert: I also want to track all merges!

Uwe

-
Uwe Schindler
H.-H.-Meier-Allee 63, D-28213 Bremen
http://www.thetaphi.de
eMail: u...@thetaphi.de


 -Original Message-
 From: Grant Ingersoll [mailto:gsing...@apache.org]
 Sent: Monday, October 18, 2010 6:48 PM
 Cc: dev@lucene.apache.org
 Subject: Re: possible to filter the output to commits@ list
 
 What commands do you want me to run?
 
 On Oct 17, 2010, at 11:27 AM, Uwe Schindler wrote:
 
  Hi Daniel,
 
  I already proposed something similar to our developers list and also
  explained, why it is a bad idea to merge single files (see attached
mail).
 
  Still our/my opinion: It may be better, to simply provide optionally
plain
  diffs without metadata for commit mails (on request of a project, e.g.
  Grant Ingersoll could configure that for the Lucene project in the SVN
  config file).
 
  Uwe
 
  -
  Uwe Schindler
  uschind...@apache.org
  Apache Lucene PMC Member / Committer
  Bremen, Germany
  http://lucene.apache.org/
 
 
  -Original Message-
  From: Daniel Shahaf [mailto:d...@daniel.shahaf.name]
  Sent: Sunday, October 17, 2010 5:17 PM
  To: Uwe Schindler
  Cc: dev@lucene.apache.org; 'Apache Infrastructure'
  Subject: Re: possible to filter the output to commits@ list
 
  Or you could try to see if some of the existing mergeinfo can be
  removed,
  and
  try to have less subtree mergeinfo in the first place; these are
  common
  topics
  on us...@subversion.
 
  Uwe Schindler wrote on Sun, Oct 17, 2010 at 17:06:44 +0200:
  Hi all,
 
  There are configuration options in SVN to let the commit mails pipe
  through a
  filter. Other svn hosting providers like Sourceforge provide some
  filters
  to be
  applied. Ideally, you would use the patchutils package (it's called
  like
  this in
  Debian/Ubuntu) and use svn diff | filterdiff --clean and pipe that
  into
  an
  eMail. Maybe we should open an issue at infra?
 
  Uwe
 
  -
  Uwe Schindler
  uschind...@apache.org
  Apache Lucene PMC Member / Committer Bremen, Germany
  http://lucene.apache.org/
 
  -Original Message-
  From: Robert Muir [mailto:rcm...@gmail.com]
  Sent: Sunday, October 17, 2010 4:58 PM
  To: dev@lucene.apache.org
  Subject: possible to filter the output to commits@ list
 
  Lets say i change a single line of code, and merge it back to the
  3x_branch.
 
  Currently we get 6 or 7 emails of mergeproperty changes to the
  commits@ list... this is making it difficult or impossible for
  backports to be reviewed at all... I think this is terrible.
 
  How can we get this fixed so that these mergeprop-changes only are
  filtered in the emails? Someone could always click the link to
  viewvc or whatever if they are interested in this...
 
  ---
  -
  - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For
  additional commands, e-mail: dev-h...@lucene.apache.org
 
 
  Mail Attachment.eml
 
 --
 Grant Ingersoll
 http://www.lucidimagination.com
 
 
 -
 To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional
 commands, e-mail: dev-h...@lucene.apache.org



-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



Re: possible to filter the output to commits@ list????

2010-10-18 Thread Grant Ingersoll

On Oct 18, 2010, at 2:07 PM, Uwe Schindler wrote:

 Hi Grant,
 
 Not yet decided/investigated, this was more a question to infra, if it would
 be possible to set this up. :-) I have to check a config from an SVN server
 that supports merging. Where are the Apach config files for SVN? Can I view
 them?

I think you will need to go through infra for the config, as even I don't have 
that.  I don't believe there is a public URL (or even PMC level)

 
 Else, I agree with Robert: I also want to track all merges!


I agree.  It was just a bit of a shocker for me the first time I did it and I 
see like 30 files changed when I only changed one file.  Once you get used to 
it, though, not so bad and the IDE definitely helps.
-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



RE: possible to filter the output to commits@ list????

2010-10-17 Thread Uwe Schindler
Hi Daniel,

I already proposed something similar to our developers list and also
explained, why it is a bad idea to merge single files (see attached mail).

Still our/my opinion: It may be better, to simply provide optionally plain
diffs without metadata for commit mails (on request of a project, e.g. Grant
Ingersoll could configure that for the Lucene project in the SVN config
file).

Uwe

-
Uwe Schindler
uschind...@apache.org 
Apache Lucene PMC Member / Committer
Bremen, Germany
http://lucene.apache.org/


 -Original Message-
 From: Daniel Shahaf [mailto:d...@daniel.shahaf.name]
 Sent: Sunday, October 17, 2010 5:17 PM
 To: Uwe Schindler
 Cc: dev@lucene.apache.org; 'Apache Infrastructure'
 Subject: Re: possible to filter the output to commits@ list
 
 Or you could try to see if some of the existing mergeinfo can be removed,
and
 try to have less subtree mergeinfo in the first place; these are common
topics
 on us...@subversion.
 
 Uwe Schindler wrote on Sun, Oct 17, 2010 at 17:06:44 +0200:
  Hi all,
 
  There are configuration options in SVN to let the commit mails pipe
through a
 filter. Other svn hosting providers like Sourceforge provide some filters
to be
 applied. Ideally, you would use the patchutils package (it's called like
this in
 Debian/Ubuntu) and use svn diff | filterdiff --clean and pipe that into
an
 eMail. Maybe we should open an issue at infra?
 
  Uwe
 
  -
  Uwe Schindler
  uschind...@apache.org
  Apache Lucene PMC Member / Committer
  Bremen, Germany
  http://lucene.apache.org/
 
   -Original Message-
   From: Robert Muir [mailto:rcm...@gmail.com]
   Sent: Sunday, October 17, 2010 4:58 PM
   To: dev@lucene.apache.org
   Subject: possible to filter the output to commits@ list
  
   Lets say i change a single line of code, and merge it back to the
3x_branch.
  
   Currently we get 6 or 7 emails of mergeproperty changes to the
   commits@ list... this is making it difficult or impossible for
   backports to be reviewed at all... I think this is terrible.
  
   How can we get this fixed so that these mergeprop-changes only are
   filtered in the emails? Someone could always click the link to
   viewvc or whatever if they are interested in this...
  
   
   - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For
   additional commands, e-mail: dev-h...@lucene.apache.org
 
 
---BeginMessage---
I cc'ed my previous eMail also to the in...@ao list. In general, if we cannot 
filter we could think about the following:

- remove all mergepros recursively for trunk, but we would lose history for 
merges.
- Alternatively merge all mergeprops up to the root folder and remove it from 
the separate files/subfolders. This would lose some changes, but if all are 
merged up, that’s fine (only some files may appear as merged from different 
branch, although they aren't).

In general when merging to reduce the amount of mergeprops:
- Don't merge single files or subdirectories! Always merge the top folder! 
Excluded from this is trunk/modules, as this needs a separate, merging step 
between trunk/3x. So a merge would then only add mergeprops to root folder and 
modules/contrib (in branch).

This is the explanation for the behavior:
SVN needs to add the mergeprops to all single files that *already* contain 
mergeprops on each merge (SVN is not able to only store diffs in per file 
props). This is the reason why we have so many mergeprops at single files that 
every time gets updated: Once a file has a merge property, it does no longer 
inherit it from the parent folder. Whenever you merge something, the new rev 
numbers get added to the parent folder, as expected; but as this single file 
also have mergeprops and those are not differential/separate to parent, the 
merge prop of this single file also needs to be updated. This leads to the 
large modification email. We cannot change that anymore, so by reducing the 
number of single-file merges, we can stop this behavior from getting worse. To 
get rid of it completely and start new, see above.

Uwe
-
Uwe Schindler
H.-H.-Meier-Allee 63, D-28213 Bremen
http://www.thetaphi.de
eMail: u...@thetaphi.de


 -Original Message-
 From: Robert Muir [mailto:rcm...@gmail.com]
 Sent: Sunday, October 17, 2010 4:58 PM
 To: dev@lucene.apache.org
 Subject: possible to filter the output to commits@ list
 
 Lets say i change a single line of code, and merge it back to the 3x_branch.
 
 Currently we get 6 or 7 emails of mergeproperty changes to the commits@
 list... this is making it difficult or impossible for backports to be 
 reviewed at
 all... I think this is terrible.
 
 How can we get this fixed so that these mergeprop-changes only are filtered in
 the emails? Someone could always click the link to viewvc or whatever if they
 are interested in this...
 
 -
 To unsubscribe, e-mail: