Re: [lang] ClassUtils.toClass(Object[] array)
null -> null makes sense. +1. On Tue, Feb 2, 2010 at 12:53 PM, Matt Benson wrote: > Currently it appears this method would throw a NPE if any element of array > is null, without documenting this fact. Our options would seem to be: > > - document the NPE > - convert a null object to Object.class > - convert a null object to a null class > > My inclination would be the third option, from the perspective of that the > expectation would be a subsequent call to one of the > ClassUtils.isAssignable(Class[]) variants. Are there dissenting opinions? > > -Matt > > - > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > > - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
Re: [math] Work going on on SVD
-- From: "Luc Maisonobe" Sent: Tuesday, February 02, 2010 11:26 AM To: "Commons Developers List" Subject: Re: [math] Work going on on SVD Luc Maisonobe a écrit : Dimitri Pourbaix a écrit : Hi, Hi Dimitri, In its present form (CM2.0), SingularValueDecomposition suffers some problems when the matrix is (numerically) singular. Luc proposed a way to improve the situation by limiting the singular values to non zero ones. Whereas the result is OK if someone is interested in getting the singular values, it is not from a purely mathematical point of view. Indeed, the resulting U matrix does no longer hold the right dimension. Instead of a number of columns equal to the number of columns of the original matrix, U now has as many columns as non-zero singular values. The product U*S*V^T yields the original matrix but the wrong size might put some users into trouble (that is true for the size of S as well). I agree with this analysis. The current behavior is interesting for some use cases but not for all of them. So I suggest we provide both cases. This could be done either by renaming the current implementation as TruncatedSVD and having your implementation named SingularValueDecomposition, or by using some constructor parameters to select the desired behavior. What do we do here ? If Dimitri can provide a new mathematically compliant SVD I would suggest to have it use the name SingularValueDecomposition and to rename the existing class TruncatedSVD. This class would take arguments to distinguish between Compact decomposition (automatic cutoff at zero value) and truncated decomposition (cutoff by singular values indices). I propose to compute U ... without taking advantage of V. That means calling EigenDecomposition a second time but should work even in case of singular matrices. That is the solution I am working on. However, doing so, I notice that EigenDecomposition also suffers major problems in case of singular matrices. A 3x3 singular matrix where 0 is an eigen value with multiplicity 2 ... yields only 2 distinct eigen vectors. The vectors associated to the null eigen value are equal!! Yes, this is JIRA issue MATH-333. The problem is that in the current implementation, the vector is computed by EigenDecomposition.findEigenVector which takes one eigenvalue as its argument. so eigenvalues with order greater than 1 simply result in the same computation repeated several times ... Perhaps one way to solve this is to reproduce what is done in DLARRV from LAPACK. The routine uses the index of the eigenvalue and not only its value. Does someone had a look at DLARRV and could explain how it works (the vector part only, the singular values part is already well understood I think) ? I think we should wait for this issue to be solved before publishing 2.1. It is currently targeted at 2.2 but I would vote to solve it earlier. It is really a big drawback of the current implementation. Not being able to decompose identity is rather sad ... +1 Having a SVD that doesn't work is really embarrassing. I'll try and find some spare cycles to look at the DLARRV implementation. However, if Dimitri can create an alternative patch, all the better. So, before I can improve SVD, I have to improve EigenDecomposition! By the way, going through SVD, EigenDecomposition, I noticed that BidiagonalTransformer and TridiagonalTransformer both use the Householder vector computation deeply imbedded in their code. In order to make both classes easier to read (and to debug), I wonder if it might be useful to introduce a class Householder which would take care of the computation of the vector in a unique place. This would be a nice improvement and would probably be useful for other linear algebra algorithms later on. The rationale for the current implementation was to avoid copying data back and forth between a low level elementary Householder class called n times and a high level transformer like bi-diagonal or tri-diagonal transformer. If we can find a way to have an elementary transform processing in-place the array provided to it by the high level transformer, this would be fine. Perhaps this could wait for 2.2 or 3.0. Luc Once again, we see we lack a way to have partial view of matrices slices. Luc Regards, Dim. Dimitri Pourbaix * Institut d'Astronomie et d'Astrophysique * Don't worry, be happy CP 226, office 2.N4.211, building NO * and CARPE DIEM. Universite Libre de Bruxelles* Boulevard du Triomphe* Tel : +32-2-650.35.71 B-1050 Bruxelles* Fax : +32-2-650.42.26 http://sb9.astro.ulb.ac.be/~pourbaix * mailto:pourb...@astro.ulb.ac.be - To unsubscribe, e-mail: dev-unsubscr...@commons.apach
[LANG] Patch attached for LANG-583 issue
Hi I have attached the patch file for LANG-583 issue in jira, Please let me know if there are any issues. Thanks Chandrashekar
Re: [lang] Preparing for a 2.5 release
On Wed, Feb 3, 2010 at 2:22 AM, Niall Pemberton wrote: > For anyone who hasn't noticed I've been back-porting Commons Lang > changes from the trunk to a 2.x branch: > > http://svn.apache.org/viewvc/commons/proper/lang/branches/LANG_2_X/ > > AFAIC its getting close to being ready for a 2.5 release. LANG-204 is > currently in progress and I will try to post a summary of whats been > back ported (and what hasn't) tomorrow. Other than that I have only > updating the site and getting the release notes & changes plugin up to > date on my todo list. > > Is there anything anyone else wants included? Currently the other Lang bugs against 2.x are the following, but I don't plan to fix them unless someone else supplies a patch: http://issues.apache.org/jira/browse/LANG-76 http://issues.apache.org/jira/browse/LANG-331 http://issues.apache.org/jira/browse/LANG-472 > Niall > - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
[lang] Preparing for a 2.5 release
For anyone who hasn't noticed I've been back-porting Commons Lang changes from the trunk to a 2.x branch: http://svn.apache.org/viewvc/commons/proper/lang/branches/LANG_2_X/ AFAIC its getting close to being ready for a 2.5 release. LANG-204 is currently in progress and I will try to post a summary of whats been back ported (and what hasn't) tomorrow. Other than that I have only updating the site and getting the release notes & changes plugin up to date on my todo list. Is there anything anyone else wants included? Niall - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
Re: [math] Work going on on SVD
Luc, I think we should wait for this issue to be solved before publishing 2.1. It is currently targeted at 2.2 but I would vote to solve it earlier. It is really a big drawback of the current implementation. Not being able to decompose identity is rather sad ... I agree. This would be a nice improvement and would probably be useful for other linear algebra algorithms later on. The rationale for the current implementation was to avoid copying data back and forth between a low level elementary Householder class called n times and a high level transformer like bi-diagonal or tri-diagonal transformer. If we can find a way to have an elementary transform processing in-place the array provided to it by the high level transformer, this would be fine. Perhaps this could wait for 2.2 or 3.0. Yes. It would be an improvement of the code from a developer perspec- tive only. No rush! Regards, Dim. Dimitri Pourbaix * Institut d'Astronomie et d'Astrophysique * Don't worry, be happy CP 226, office 2.N4.211, building NO * and CARPE DIEM. Universite Libre de Bruxelles* Boulevard du Triomphe* Tel : +32-2-650.35.71 B-1050 Bruxelles* Fax : +32-2-650.42.26 http://sb9.astro.ulb.ac.be/~pourbaix * mailto:pourb...@astro.ulb.ac.be - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
[lang] ClassUtils.toClass(Object[] array)
Currently it appears this method would throw a NPE if any element of array is null, without documenting this fact. Our options would seem to be: - document the NPE - convert a null object to Object.class - convert a null object to a null class My inclination would be the third option, from the perspective of that the expectation would be a subsequent call to one of the ClassUtils.isAssignable(Class[]) variants. Are there dissenting opinions? -Matt - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
Re: [math] Work going on on SVD
Luc Maisonobe a écrit : > Dimitri Pourbaix a écrit : >> Hi, > > Hi Dimitri, > >> In its present form (CM2.0), SingularValueDecomposition suffers some >> problems when the matrix is (numerically) singular. Luc proposed a >> way to improve the situation by limiting the singular values to non >> zero ones. Whereas the result is OK if someone is interested in getting >> the singular values, it is not from a purely mathematical point of view. >> Indeed, the resulting U matrix does no longer hold the right dimension. >> Instead of a number of columns equal to the number of columns of the >> original matrix, U now has as many columns as non-zero singular values. >> The product U*S*V^T yields the original matrix but the wrong size might >> put some users into trouble (that is true for the size of S as well). > > I agree with this analysis. The current behavior is interesting for some > use cases but not for all of them. So I suggest we provide both cases. > This could be done either by renaming the current implementation as > TruncatedSVD and having your implementation named > SingularValueDecomposition, or by using some constructor parameters to > select the desired behavior. What do we do here ? If Dimitri can provide a new mathematically compliant SVD I would suggest to have it use the name SingularValueDecomposition and to rename the existing class TruncatedSVD. This class would take arguments to distinguish between Compact decomposition (automatic cutoff at zero value) and truncated decomposition (cutoff by singular values indices). > >> I propose to compute U ... without taking advantage of V. That means >> calling EigenDecomposition a second time but should work even in case >> of singular matrices. That is the solution I am working on. However, >> doing so, I notice that EigenDecomposition also suffers major problems >> in case of singular matrices. A 3x3 singular matrix where 0 is an >> eigen value with multiplicity 2 ... yields only 2 distinct eigen >> vectors. The vectors associated to the null eigen value are equal!! > > Yes, this is JIRA issue MATH-333. The problem is that in the current > implementation, the vector is computed by > EigenDecomposition.findEigenVector which takes one eigenvalue as its > argument. so eigenvalues with order greater than 1 simply result in the > same computation repeated several times ... > Perhaps one way to solve this is to reproduce what is done in DLARRV > from LAPACK. The routine uses the index of the eigenvalue and not only > its value. Does someone had a look at DLARRV and could explain how it works (the vector part only, the singular values part is already well understood I think) ? I think we should wait for this issue to be solved before publishing 2.1. It is currently targeted at 2.2 but I would vote to solve it earlier. It is really a big drawback of the current implementation. Not being able to decompose identity is rather sad ... > >> So, before I can improve SVD, I have to improve EigenDecomposition! >> >> By the way, going through SVD, EigenDecomposition, I noticed that >> BidiagonalTransformer and TridiagonalTransformer both use the >> Householder vector computation deeply imbedded in their code. In >> order to make both classes easier to read (and to debug), I wonder >> if it might be useful to introduce a class Householder which would >> take care of the computation of the vector in a unique place. > > This would be a nice improvement and would probably be useful for other > linear algebra algorithms later on. The rationale for the current > implementation was to avoid copying data back and forth between a low > level elementary Householder class called n times and a high level > transformer like bi-diagonal or tri-diagonal transformer. If we can find > a way to have an elementary transform processing in-place the array > provided to it by the high level transformer, this would be fine. Perhaps this could wait for 2.2 or 3.0. Luc > > Once again, we see we lack a way to have partial view of matrices slices. > > Luc > > >> Regards, >> Dim. >> >> >> Dimitri Pourbaix * >> Institut d'Astronomie et d'Astrophysique * Don't worry, be happy >> CP 226, office 2.N4.211, building NO * and CARPE DIEM. >> Universite Libre de Bruxelles* >> Boulevard du Triomphe* Tel : +32-2-650.35.71 >> B-1050 Bruxelles* Fax : +32-2-650.42.26 >> http://sb9.astro.ulb.ac.be/~pourbaix * mailto:pourb...@astro.ulb.ac.be >> >> - >> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >> For additional commands, e-mail: dev-h...@commons.apache.org >> > > > - > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional
Re: svn commit: r904829 - in /commons/proper/lang/branches/LANG_2_X/src: main/java/org/apache/commons/lang/time/DateUtils.java test/java/org/apache/commons/lang/time/DateUtilsTest.java
On Tue, Feb 2, 2010 at 5:32 PM, sebb wrote: > On 02/02/2010, sebb wrote: >> On 30/01/2010, nia...@apache.org wrote: >> > Author: niallp >> > Date: Sat Jan 30 17:49:28 2010 >> > New Revision: 904829 >> > >> > URL: http://svn.apache.org/viewvc?rev=904829&view=rev >> > Log: >> > Port LANG-530 to 2.x branch - Fix parseDate() cannot parse ISO8601 dates >> produced by FastDateFormat >> > >> > Modified: >> >> >> > >> > commons/proper/lang/branches/LANG_2_X/src/test/java/org/apache/commons/lang/time/DateUtilsTest.java >> > >> >> >> > + // http://issues.apache.org/jira/browse/LANG-520 >> > + public void testLang520() throws ParseException { >> >> >> Looks like the above test and comment should really be LANG-530 >> (likewise in lang3)? >> > > I've just discovered that the test fails with a ParseException on Java > 1.3 - this is because SimpleDateFormat does not support 'Z' in that > version of Java. I'll add an exclusion. Thanks Niall >> S/// >> > > - > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > > - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
Re: svn commit: r904829 - in /commons/proper/lang/branches/LANG_2_X/src: main/java/org/apache/commons/lang/time/DateUtils.java test/java/org/apache/commons/lang/time/DateUtilsTest.java
On 02/02/2010, sebb wrote: > On 30/01/2010, nia...@apache.org wrote: > > Author: niallp > > Date: Sat Jan 30 17:49:28 2010 > > New Revision: 904829 > > > > URL: http://svn.apache.org/viewvc?rev=904829&view=rev > > Log: > > Port LANG-530 to 2.x branch - Fix parseDate() cannot parse ISO8601 dates > produced by FastDateFormat > > > > Modified: > > > > > > commons/proper/lang/branches/LANG_2_X/src/test/java/org/apache/commons/lang/time/DateUtilsTest.java > > > > > > +// http://issues.apache.org/jira/browse/LANG-520 > > +public void testLang520() throws ParseException { > > > Looks like the above test and comment should really be LANG-530 > (likewise in lang3)? > I've just discovered that the test fails with a ParseException on Java 1.3 - this is because SimpleDateFormat does not support 'Z' in that version of Java. I'll add an exclusion. > S/// > - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
Re: svn commit: r904829 - in /commons/proper/lang/branches/LANG_2_X/src: main/java/org/apache/commons/lang/time/DateUtils.java test/java/org/apache/commons/lang/time/DateUtilsTest.java
On 30/01/2010, nia...@apache.org wrote: > Author: niallp > Date: Sat Jan 30 17:49:28 2010 > New Revision: 904829 > > URL: http://svn.apache.org/viewvc?rev=904829&view=rev > Log: > Port LANG-530 to 2.x branch - Fix parseDate() cannot parse ISO8601 dates > produced by FastDateFormat > > Modified: > > commons/proper/lang/branches/LANG_2_X/src/test/java/org/apache/commons/lang/time/DateUtilsTest.java > > +// http://issues.apache.org/jira/browse/LANG-520 > +public void testLang520() throws ParseException { Looks like the above test and comment should really be LANG-530 (likewise in lang3)? S/// - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
[Commons Wiki] Update of "CreatingReleases" by HenriBie stro
Dear Wiki user, You have subscribed to a wiki page or wiki category on "Commons Wiki" for change notification. The "CreatingReleases" page has been changed by HenriBiestro. http://wiki.apache.org/commons/CreatingReleases?action=diff&rev1=41&rev2=42 -- E.3 Deploy the Site Run {{{mvn site-deploy}}} to deploy the site - please note that you are deploying the site of the next development snapshot. + If your local username is not the same as your Apache username, run: {{{mvn -Duser.name=apachename site-deploy}}}. E.4 Create the SVN Release Tag - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
[RESULT] Release Jexl 2.0
The following people voted on Jexl 2.0 release: Rahul Akolkar: +1 Oliver Heger: +1 Seb Baz : +1 Phil Steiz: +1 Jorg Schaible: +1 Thanks to all of you. I'll cut the 2.0 release in the next couple of days. Cheers; Henrib -- View this message in context: http://n4.nabble.com/RESULT-Release-Jexl-2-0-tp1460048p1460048.html Sent from the Commons - Dev mailing list archive at Nabble.com. - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
Re: [lang] Working on issue LANG - 583 Was: Need to start the work
Sir: On 2 February 2010 00:30, Chandra Shekar wrote: > ok i can download the code, But how can i asssign the issue to myself in > jira, I dont see any option for assign You don't assign yourself a particular issue. You publish the patch to the issue, then some comitter evaluates it for inclusion at which point it is either accepted or not. http://commons.apache.org/math/developers.html is the Commons-Math commit process. Other projects within the commons hierachy have similar processes, and I'm sure someone else will jump in and point out any differences. -- Sent from my mobile device Envoyait de mon telephone mobil - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
Re: Need to start the work
You can download the code through svn (see http://commons.apache.org/svninfo.html), and then you just have to attach the patch to the bug report. Valentin. Le 02/02/2010 08:10, Chandra Shekar a écrit : Hi I want to assign the bug LANG - 583 to myself, I am not able to see asign option, also i want to know where i can download the code work and then submit the patch, can any body help me on this. Thanks Chandrashekar - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
Re: [lang] Working on issue LANG - 583 Was: Need to start the work
One More thing, ichecked out the code using the following command from my netbeans ide directly http://svn.apache.org/repos/asf/commons/proper/lang/trunk i am getting some compile time errors in the checked out code. Thanks Chandrashekar --- On Tue, 2/2/10, Chandra Shekar wrote: From: Chandra Shekar Subject: Re: [lang] Working on issue LANG - 583 Was: Need to start the work To: "Commons Developers List" Date: Tuesday, February 2, 2010, 8:30 AM ok i can download the code, But how can i asssign the issue to myself in jira, I dont see any option for assign Thanks Chandrashekar --- On Tue, 2/2/10, Julien Aymé wrote: From: Julien Aymé Subject: Re: [lang] Working on issue LANG - 583 Was: Need to start the work To: "Commons Developers List" Date: Tuesday, February 2, 2010, 7:27 AM Hello Chandra, you should prefix your mail object with [lang] to have a faster feedback, since most of the commons developers have rules that match the component name. In order to work on this issue you should check out the trunk version of lang (see http://commons.apache.org/lang/source-repository.html), the you can work on your local copy and attach a patch to the JIRA issue. Regards, Julien 2010/2/2 Chandra Shekar : > Hi > > I want to assign the bug LANG - 583 to myself, I am not able to see asign > option, also i want to know where i can download the code work and then > submit the patch, > > can any body help me on this. > > Thanks > Chandrashekar > > > > > - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
Re: [lang] Working on issue LANG - 583 Was: Need to start the work
ok i can download the code, But how can i asssign the issue to myself in jira, I dont see any option for assign Thanks Chandrashekar --- On Tue, 2/2/10, Julien Aymé wrote: From: Julien Aymé Subject: Re: [lang] Working on issue LANG - 583 Was: Need to start the work To: "Commons Developers List" Date: Tuesday, February 2, 2010, 7:27 AM Hello Chandra, you should prefix your mail object with [lang] to have a faster feedback, since most of the commons developers have rules that match the component name. In order to work on this issue you should check out the trunk version of lang (see http://commons.apache.org/lang/source-repository.html), the you can work on your local copy and attach a patch to the JIRA issue. Regards, Julien 2010/2/2 Chandra Shekar : > Hi > > I want to assign the bug LANG - 583 to myself, I am not able to see asign > option, also i want to know where i can download the code work and then > submit the patch, > > can any body help me on this. > > Thanks > Chandrashekar > > > > > - To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org