Re: RFR: 8283994: Make Xerces DatatypeException stackless [v2]
On Wed, 13 Apr 2022 08:30:33 GMT, Aleksey Shipilev wrote: >>> Hello @shipilev, do you think this stackless nature of this specific >>> `DatatypeException` type should be noted in its javadoc, just to avoid any >>> surprises when someone in future ends up using this exception type as the >>> "cause" of some other exception? >> >> I don't think so. It seems to me the intent for these exceptions is to carry >> error information without any stack trace info. > >> @shipilev >> >> > Any other reviews? I would like someone else to confirm my investigation >> > that we don't use the stack traces out of these exceptions too... >> >> I opened [XERCESJ‑1742](https://issues.apache.org/jira/browse/XERCESJ-1742) >> to get this change upstreamed into **Xerces**, the resolution of which >> should also answer your question about whether the stack traces are really >> unused. > > Thanks! But I see there seem to be no upstream interest yet. So I am planning > to integrate this to JDK first, and so asking for JDK-specific reviews again. > > Hello @shipilev, do you think this stackless nature of this specific > > `DatatypeException` type should be noted in its javadoc, just to avoid any > > surprises when someone in future ends up using this exception type as the > > "cause" of some other exception? > > I don't think so. It seems to me the intent for these exceptions is to carry > error information without any stack trace info. Looks fine to me then. However, I'm not a Reviewer. - PR: https://git.openjdk.java.net/jdk/pull/8036
Re: RFR: 8283994: Make Xerces DatatypeException stackless [v2]
On Thu, 31 Mar 2022 17:54:28 GMT, Aleksey Shipilev wrote: >> See bug report for more details. This change improves >> SPECjvm2008:xml.validation for about +3%: >> >> >> baseline: 298.353 ± 1.008 ops/min >> patched: 309.912 ± 1.347 ops/min >> >> Of course, the real improvements might be even higher, as exception might be >> thrown from much deeper call hierarchy. >> >> Additional testing: >> - [x] Linux x86_64 fastdebug, `jaxp_all` >> - [x] Linux x86_64 fastdebug, `javax/xml` > > Aleksey Shipilev has updated the pull request incrementally with one > additional commit since the last revision: > > Also update LastModified Marked as reviewed by jpai (Committer). - PR: https://git.openjdk.java.net/jdk/pull/8036
Re: RFR: 8283994: Make Xerces DatatypeException stackless [v2]
On Wed, 6 Apr 2022 07:48:04 GMT, Aleksey Shipilev wrote: >> Any other reviews? I would like someone else to confirm my investigation >> that we don't use the stack traces out of these exceptions too... > >> Hello @shipilev, do you think this stackless nature of this specific >> `DatatypeException` type should be noted in its javadoc, just to avoid any >> surprises when someone in future ends up using this exception type as the >> "cause" of some other exception? > > I don't think so. It seems to me the intent for these exceptions is to carry > error information without any stack trace info. > @shipilev > > > Any other reviews? I would like someone else to confirm my investigation > > that we don't use the stack traces out of these exceptions too... > > I opened [XERCESJ‑1742](https://issues.apache.org/jira/browse/XERCESJ-1742) > to get this change upstreamed into **Xerces**, the resolution of which should > also answer your question about whether the stack traces are really unused. Thanks! But I see there seem to be no upstream interest yet. So I am planning to integrate this to JDK first, and so asking for JDK-specific reviews again. - PR: https://git.openjdk.java.net/jdk/pull/8036
Re: RFR: 8283994: Make Xerces DatatypeException stackless [v2]
On Fri, 1 Apr 2022 08:10:48 GMT, Aleksey Shipilev wrote: >> Aleksey Shipilev has updated the pull request incrementally with one >> additional commit since the last revision: >> >> Also update LastModified > > Any other reviews? I would like someone else to confirm my investigation that > we don't use the stack traces out of these exceptions too... > Hello @shipilev, do you think this stackless nature of this specific > `DatatypeException` type should be noted in its javadoc, just to avoid any > surprises when someone in future ends up using this exception type as the > "cause" of some other exception? I don't think so. It seems to me the intent for these exceptions is to carry error information without any stack trace info. - PR: https://git.openjdk.java.net/jdk/pull/8036
Re: RFR: 8283994: Make Xerces DatatypeException stackless
On Wed, 30 Mar 2022 11:38:59 GMT, Aleksey Shipilev wrote: > See bug report for more details. This change improves > SPECjvm2008:xml.validation for about +3%: > > > baseline: 298.353 ± 1.008 ops/min > patched: 309.912 ± 1.347 ops/min > > Of course, the real improvements might be even higher, as exception might be > thrown from much deeper call hierarchy. > > Additional testing: > - [x] Linux x86_64 fastdebug, `jaxp_all` > - [x] Linux x86_64 fastdebug, `javax/xml` Marked as reviewed by joehw (Reviewer). src/java.xml/share/classes/com/sun/org/apache/xerces/internal/impl/dv/DatatypeException.java line 2: > 1: /* > 2: * Copyright (c) 2017, 2022, Oracle and/or its affiliates. All rights > reserved. Update the LastModified tag below as well. Otherwise, looks good. - PR: https://git.openjdk.java.net/jdk/pull/8036