Menu
▾
▴
jython-dev
|
From: Jeff A. <ja...@fa...> - 2019-10-06 09:56:30
|
I think the version of Jython now at the repo tip is, apart from changing the metadata, a viable Jython 2.7.2 beta. I have dry run the process several times to the point where I can make the artefacts that would go to Sonatype, at least as far as I can test that. I have ironed out the problems (as they arise on my machine) and worked on the ant scripts. I think it's a straight run now for Frank to build and publish. The process is here: https://jython-devguide.readthedocs.io/en/latest/release_jy.html I have bumped Jython up to Java 8. This broke a few things, now resolved. JarJar does not translate the later bytecode in the version we had used. (See issue 2806.) And Java 8 is picky about Javadoc, where apparently we've not been good :o(. I fixed over a hundred javadoc errors, but I'm leaving the warnings for a later burst of enthusiasm. A lot of bugs are fixed in this release and some features added as noted in NEWS. I still have misgivings around the number of suppressed tests (and so many places we can do it), and the way tests involving networking (SSL and Netty) contiunue to be flakey. The socket-related modules are an awesome achievement, but incomplete in some ways. However, one must not let the defects dominate one's thinking. So much is good and improved in 2.7.2, and this includes all the issues we prioritised in the triage. What do others think? Tag it? Jeff -- Jeff Allen |
|
From: Jim B. <jim...@py...> - 2019-10-06 17:15:10
|
Jeff, I'm glad to hear this great news! We should proceed now as you suggest, given that you have reached the far as you can tell point in the testing process. So let's use this beta to shake out the release process. At this point in this release cycle, our code quality and corresponding testing is very good, so it should not be considered an alpha. Future releases can build further and also provide more assurance on shrinking the time to subsequent releases. Gradle support is a good example here. One nit: there's still an obsolete reference to Java 7 in https://jython-devguide.readthedocs.io/en/latest/release_jy.html : "At the time of writing we target Java 7." - Jim On Sun, Oct 6, 2019 at 2:57 AM Jeff Allen <ja...@fa...> wrote: > I think the version of Jython now at the repo tip is, apart from changing > the metadata, a viable Jython 2.7.2 beta. > > I have dry run the process several times to the point where I can make the > artefacts that would go to Sonatype, at least as far as I can test that. I > have ironed out the problems (as they arise on my machine) and worked on > the ant scripts. I think it's a straight run now for Frank to build and > publish. > > The process is here: > https://jython-devguide.readthedocs.io/en/latest/release_jy.html > > I have bumped Jython up to Java 8. This broke a few things, now resolved. > JarJar does not translate the later bytecode in the version we had used. > (See issue 2806.) And Java 8 is picky about Javadoc, where apparently we've > not been good :o(. I fixed over a hundred javadoc errors, but I'm leaving > the warnings for a later burst of enthusiasm. > > A lot of bugs are fixed in this release and some features added as noted > in NEWS. I still have misgivings around the number of suppressed tests (and > so many places we can do it), and the way tests involving networking (SSL > and Netty) contiunue to be flakey. The socket-related modules are an > awesome achievement, but incomplete in some ways. However, one must not let > the defects dominate one's thinking. So much is good and improved in 2.7.2, > and this includes all the issues we prioritised in the triage. > > What do others think? Tag it? > > Jeff > > -- > Jeff Allen > > _______________________________________________ > Jython-dev mailing list > Jyt...@li... > https://lists.sourceforge.net/lists/listinfo/jython-dev > |
|
From: Jeff A. <ja...@fa...> - 2019-10-06 19:59:32
|
On 06/10/2019 17:52, Jim Baker wrote: > Jeff, > > I'm glad to hear this great news! We should proceed now as you > suggest, given that you have reached the far as you can tell point in > the testing process. I doubt anyone will disagree, but I thought I'd give a couple of days for others to respond. > > So let's use this beta to shake out the release process. At this point > in this release cycle, our code quality and corresponding testing is > very good, so it should not be considered an alpha. Future releases > can build further and also provide more assurance on shrinking the > time to subsequent releases. Gradle support is a good example here. > > One nit: there's still an obsolete reference to Java 7 in > https://jython-devguide.readthedocs.io/en/latest/release_jy.html : > "At the time of writing we target Java 7." > Ah yes. And in the following code display too. Thanks for reading carefully. I'll fix that. Jeff |
|
From: <fwi...@gm...> - 2019-10-06 19:34:32
|
On Sun, Oct 6, 2019 at 2:56 AM Jeff Allen <ja...@fa...> wrote: > > I think the version of Jython now at the repo tip is, apart from changing the metadata, a viable Jython 2.7.2 beta. > > I have dry run the process several times to the point where I can make the artefacts that would go to Sonatype, at least as far as I can test that. I have ironed out the problems (as they arise on my machine) and worked on the ant scripts. I think it's a straight run now for Frank to build and publish. > > The process is here: https://jython-devguide.readthedocs.io/en/latest/release_jy.html When I build and upload I'll see if these docs need updating. > > I have bumped Jython up to Java 8. This broke a few things, now resolved. JarJar does not translate the later bytecode in the version we had used. (See issue 2806.) And Java 8 is picky about Javadoc, where apparently we've not been good :o(. I fixed over a hundred javadoc errors, but I'm leaving the warnings for a later burst of enthusiasm. > > A lot of bugs are fixed in this release and some features added as noted in NEWS. I still have misgivings around the number of suppressed tests (and so many places we can do it), and the way tests involving networking (SSL and Netty) contiunue to be flakey. The socket-related modules are an awesome achievement, but incomplete in some ways. However, one must not let the defects dominate one's thinking. So much is good and improved in 2.7.2, and this includes all the issues we prioritised in the triage. > This is great news! > What do others think? Tag it? Yes, tag it! I agree with Jim that this should be tagged as a beta release. I confirmed that my sonatype login still works, I'm happy to test out the release process and get this published. While I'm there I'll poke around to see if I can find out how new users are added. > Jeff > > -- > Jeff Allen |
|
From: Jeff A. <ja...@fa...> - 2019-10-08 19:59:11
|
Ok, let's do it! I have pushed the changesets I made when I did the last dry run (the metadata changes and the tag). This means that in the documented process, sections 23.1.3, 23.1.6 and 23.1.10 are already taken care of. I did the tests leading up to 23.1.10, but feel free to repeat enough of them to protect our reputation. This full build has only ever run on a Windows machine, so far as I know, so there is definitely scope for latent bugs. When it comes to building the bundles, I think I reproduced what was there before, but it is entirely untested by me. There is a new bundle for jython-slim. There are sections of |maven/build.xml| that looked like contingent work-arounds and I'm unable to fix/test the ones that involve the bundles. Good luck! Jeff Allen On 06/10/2019 20:33, fwi...@gm... wrote: > On Sun, Oct 6, 2019 at 2:56 AM Jeff Allen <ja...@fa...> wrote: >> I think the version of Jython now at the repo tip is, apart from changing the metadata, a viable Jython 2.7.2 beta. >> >> I have dry run the process several times to the point where I can make the artefacts that would go to Sonatype, at least as far as I can test that. I have ironed out the problems (as they arise on my machine) and worked on the ant scripts. I think it's a straight run now for Frank to build and publish. >> >> The process is here: https://jython-devguide.readthedocs.io/en/latest/release_jy.html > When I build and upload I'll see if these docs need updating. > >> I have bumped Jython up to Java 8. This broke a few things, now resolved. JarJar does not translate the later bytecode in the version we had used. (See issue 2806.) And Java 8 is picky about Javadoc, where apparently we've not been good :o(. I fixed over a hundred javadoc errors, but I'm leaving the warnings for a later burst of enthusiasm. >> >> A lot of bugs are fixed in this release and some features added as noted in NEWS. I still have misgivings around the number of suppressed tests (and so many places we can do it), and the way tests involving networking (SSL and Netty) contiunue to be flakey. The socket-related modules are an awesome achievement, but incomplete in some ways. However, one must not let the defects dominate one's thinking. So much is good and improved in 2.7.2, and this includes all the issues we prioritised in the triage. >> > This is great news! > >> What do others think? Tag it? > Yes, tag it! I agree with Jim that this should be tagged as a beta > release. I confirmed that my sonatype login still works, I'm happy to > test out the release process and get this published. While I'm there > I'll poke around to see if I can find out how new users are added. > >> Jeff >> >> -- >> Jeff Allen |
|
From: <fwi...@gm...> - 2019-10-09 05:42:54
|
On Tue, Oct 8, 2019 at 12:58 PM Jeff Allen <ja...@fa...> wrote: > > Ok, let's do it! Great! I have started giving it a try locally but have run into a few problems. I'm pretty sure all of the problems are on me - I will reach out if I think I need help. I'll keep trying tonight but I think getting my cobwebs out of my environment is probably going to take more than one day. I will keep you posted. -Frank |
|
From: <fwi...@gm...> - 2019-10-10 05:26:48
|
OK - I got my environment issues worked out (some dumb stuff with my local brew setup and mercurial), and I was able to follow the release procedure here: https://jython-devguide.readthedocs.io/en/latest/release_jy.html I think the release process and docs are in good shape. I avoided looking at the gradle stuff for now :) In section 23.1.7, it suggests that that I run `ant full-check` and later suggests that this will create a build, but I think we want `ant full-build`, does that sound right? I'm rusty so I could be missing something. On OSX I get one extra failure on `test_socket_jy` but it looks similar to the test_ssl_jy failure so not a big deal IMO. I think it looks like a step was removed from maven/build.xml - it previously gpg signed the bundles to be uploaded to maven. I think this step is necessary to publish the bundles. I'll look into it in more detail tomorrow. But, progress - almost there! BTW when I get this all published, would you want to do the announcement Jeff? I'm happy to do it, but I'm also happy to just point to your announcement if you would like to do it. -Frank |
|
From: Jeff A. <ja...@fa...> - 2019-10-10 08:47:37
|
On 10/10/2019 06:26, fwi...@gm... wrote: > OK - I got my environment issues worked out (some dumb stuff with my > local brew setup and mercurial), and I was able to follow the release > procedure here: > https://jython-devguide.readthedocs.io/en/latest/release_jy.html I > think the release process and docs are in good shape. I avoided > looking at the gradle stuff for now :) Thanks for diving in. If I understand correctly, you don't need to install Gradle because .\gradlew (w is for wrapper, not Windows) and its subdirectory brings all you need. > In section 23.1.7, it suggests that that I run `ant full-check` and > later suggests that this will create a build, but I think we want `ant > full-build`, does that sound right? I'm rusty so I could be missing > something. In 23.1.7 we run full-check again as a safety check that we're actually at the tag, that the output prints what you see in the display box. The build does various integrity checks. In the subsequent paragraph (text), it says to run the full-build. Could be more explicit. > On OSX I get one extra failure on `test_socket_jy` but it looks > similar to the test_ssl_jy failure so not a big deal IMO. I find networking tests have to be taken with a pinch of salt on personal devices, I think because of their interaction with security products or ISA DNS hijacking. OTOH networking is an area where Jython hangs by a thread (or many concurrent threads :). I take some comfort that this ran ok on CI: https://travis-ci.org/jythontools/jython/jobs/595290735#L647 > I think it looks like a step was removed from maven/build.xml - it > previously gpg signed the bundles to be uploaded to maven. I think > this step is necessary to publish the bundles. I'll look into it in > more detail tomorrow. Yes, it failed gracelessly on my machine and looked optional (https://hg.python.org/jython/file/05f61ba4d447/maven/build.xml#l172). Presumably the person using the script has to have a (your) private key anyway for the result to be acceptable. It looks like the signatures will sit alongside the JAR files as separate files, but I don't know the tool. Sorry, I should have created a personal account at Sonatype to dry-run this. Could it be done manually at the prompt for now? Note that the location of the bundles is ./publications now and that's all there is in it, so easy to glob. If you edit maven/build.xml the safeguards will force you to make a snapshot becuase you have changes not checked-in at the tag. I have raised issue 2814 to fix. > But, progress - almost there! > > BTW when I get this all published, would you want to do the > announcement Jeff? I'm happy to do it, but I'm also happy to just > point to your announcement if you would like to do it. You have various channels under your control: I don't see why you wouldn't use those with much the same announcement as before. I'm able to update the website. (I think we want to show 2.7.1 as current and the beta alongside.) It's adveretising. Let's do everything. All: Any thoughts on target duration? I think maybe a fixed 6 weeks (subject to personal commitments) and the next thing is either b2 or rc1 depending on how many new bugs appear and their criticality. Jeff |
|
From: <fwi...@gm...> - 2019-10-11 06:04:37
|
I have the bundles built and gpg signed, but I'm still missing something that prevents me from finalizing the uploads, I'll keep trying and post here when we have soft releases available. > You have various channels under your control: I don't see why you wouldn't use those with much the same announcement as before. I'm able to update the website. (I think we want to show 2.7.1 as current and the beta alongside.) It's adveretising. Let's do everything. > Sounds good, when I get the release uploaded I will do the announcements as usual. > All: Any thoughts on target duration? I think maybe a fixed 6 weeks (subject to personal commitments) and the next thing is either b2 or rc1 depending on how many new bugs appear and their criticality. 6 weeks works for me. -Frank |
|
From: Jeff A. <ja...@fa...> - 2019-10-11 07:00:44
|
On 11/10/2019 07:03, fwi...@gm... wrote: > I have the bundles built and gpg signed, but I'm still missing > something that prevents me from finalizing the uploads, I'll keep > trying and post here when we have soft releases available. It's disappointing the upload tool (website I assume) doesn't pinpoint the problem. I think I'd misunderstood earlier: we sign JARs and POM individually and bundle don't we, rather than sign each bundle? So where my build stops (with 4 purported bundles) isn't a feasible starting point: the problem is further back, and these "bundles" don't have signatures for the objects inside. So (explaining to myself), it's the contents of ./build/maven that need (individually) signing, as generated by the targets bundle, bundle-standalone, etc.. The maven/build.xml we inherited is this: https://hg.python.org/jython/file/3b424500eab9/maven/build.xml and my changes have been to eliminate the hard-coding of project.version and to include the Gradle-built JAR if it exists. I also changed build.dir to build.maven (=./build/maven), because the same name meaning something completely different from in the main build was confusing me. I thought it odd that the hard-coding was for 2.7.0 beta, whereas I know we have published several times since then. This suggests we didn't inherit the latest version. If this gets too tangled, we could suspend the attempt and I'll try to develop (and encapsulate) a working publication process, tested against an account I create and own. But Sonatype confuses me: I haven't even found the place it invites me to make such an account. Jeff >> You have various channels under your control: I don't see why you wouldn't use those with much the same announcement as before. I'm able to update the website. (I think we want to show 2.7.1 as current and the beta alongside.) It's adveretising. Let's do everything. >> > Sounds good, when I get the release uploaded I will do the > announcements as usual. > >> All: Any thoughts on target duration? I think maybe a fixed 6 weeks (subject to personal commitments) and the next thing is either b2 or rc1 depending on how many new bugs appear and their criticality. > 6 weeks works for me. > > -Frank > |
|
From: Adam B. <ada...@gm...> - 2019-10-11 07:17:05
|
+1 on a six week beta Adam > 在 2019年10月11日，下午5:01，Jeff Allen <ja...@fa...> 写道： > > On 11/10/2019 07:03, fwi...@gm... wrote: >> I have the bundles built and gpg signed, but I'm still missing >> something that prevents me from finalizing the uploads, I'll keep >> trying and post here when we have soft releases available. > It's disappointing the upload tool (website I assume) doesn't pinpoint the problem. I think I'd misunderstood earlier: we sign JARs and POM individually and bundle don't we, rather than sign each bundle? So where my build stops (with 4 purported bundles) isn't a feasible starting point: the problem is further back, and these "bundles" don't have signatures for the objects inside. So (explaining to myself), it's the contents of ./build/maven that need (individually) signing, as generated by the targets bundle, bundle-standalone, etc.. > > The maven/build.xml we inherited is this: https://hg.python.org/jython/file/3b424500eab9/maven/build.xml and my changes have been to eliminate the hard-coding of project.version and to include the Gradle-built JAR if it exists. I also changed build.dir to build.maven (=./build/maven), because the same name meaning something completely different from in the main build was confusing me. > > I thought it odd that the hard-coding was for 2.7.0 beta, whereas I know we have published several times since then. This suggests we didn't inherit the latest version. > > If this gets too tangled, we could suspend the attempt and I'll try to develop (and encapsulate) a working publication process, tested against an account I create and own. But Sonatype confuses me: I haven't even found the place it invites me to make such an account. > > Jeff > >>> You have various channels under your control: I don't see why you wouldn't use those with much the same announcement as before. I'm able to update the website. (I think we want to show 2.7.1 as current and the beta alongside.) It's adveretising. Let's do everything. >>> >> Sounds good, when I get the release uploaded I will do the >> announcements as usual. >> >>> All: Any thoughts on target duration? I think maybe a fixed 6 weeks (subject to personal commitments) and the next thing is either b2 or rc1 depending on how many new bugs appear and their criticality. >> 6 weeks works for me. >> >> -Frank >> > _______________________________________________ > Jython-dev mailing list > Jyt...@li... > https://lists.sourceforge.net/lists/listinfo/jython-dev |
|
From: Jeff A. <ja...@fa...> - 2019-10-12 21:19:30
|
Hi Frank: Evidently this is keeping you up late. If it hasn't suddenly come right for you, I think I should take this back: seems we're not ready. I've got myself GnuPG. I think I could dry run the whole process, fixing my broken scripts as I go, as I did with the rest.I think the outcome would now have to be tagged v2.7.2b2 to avoid confusion. A parallel possibility is to post v2.7.2b1 somewhere and just link it from the website. Jeff On 11/10/2019 07:03, fwi...@gm... wrote: > I have the bundles built and gpg signed, but I'm still missing > something that prevents me from finalizing the uploads, I'll keep > trying and post here when we have soft releases available. > >> You have various channels under your control: I don't see why you wouldn't use those with much the same announcement as before. I'm able to update the website. (I think we want to show 2.7.1 as current and the beta alongside.) It's adveretising. Let's do everything. >> > Sounds good, when I get the release uploaded I will do the > announcements as usual. > >> All: Any thoughts on target duration? I think maybe a fixed 6 weeks (subject to personal commitments) and the next thing is either b2 or rc1 depending on how many new bugs appear and their criticality. > 6 weeks works for me. > > -Frank > |
|
From: <fwi...@gm...> - 2019-10-13 04:40:54
|
On Sat, Oct 12, 2019 at 2:19 PM Jeff Allen <ja...@fa...> wrote: > > Hi Frank: > > Evidently this is keeping you up late. If it hasn't suddenly come right for you, I think I should take this back: seems we're not ready. No worries! Since I'm the only one that has released Jython in the last decade or so, I'm sure there is something peculiar in my environment that is interacting poorly with the current changes. I bet if I grab an old version of build.xml I may be able to get the process to work. I don't have time tonight but I'm sure I can give it a try tomorrow. > I've got myself GnuPG. I think I could dry run the whole process, fixing my broken scripts as I go, as I did with the rest.I think the outcome would now have to be tagged v2.7.2b2 to avoid confusion. A parallel possibility is to post v2.7.2b1 somewhere and just link it from the website. > A v2.2.2b2 sounds perfectly fine to me if it turns out that I'm unable to get the current version working, but I do want to give it another try. Once I figure out what is needed I'll make sure to let everyone know what the fix was. -Frank |
|
From: Jeff A. <ja...@fa...> - 2019-10-13 08:20:33
|
On 13/10/2019 05:40, fwi...@gm... wrote: > On Sat, Oct 12, 2019 at 2:19 PM Jeff Allen <ja...@fa...> wrote: >> Hi Frank: >> >> Evidently this is keeping you up late. If it hasn't suddenly come right for you, I think I should take this back: seems we're not ready. > No worries! Since I'm the only one that has released Jython in the > last decade or so, I'm sure there is something peculiar in my > environment that is interacting poorly with the current changes. I bet > if I grab an old version of build.xml I may be able to get the process > to work. I don't have time tonight but I'm sure I can give it a try > tomorrow. Ok, well let us know when you've had enough :). Meanwhile I'll learn what I can about the process. The video <https://youtu.be/dXR4pJ_zS-0?t=300> suggests that when it fails there's quite a lot of evidence where the bundle contents are defective. An old ./maven/build.xml is worth trying. I'm confident the current main ./build.xml delivers the right things. The safeguards are all in the main one (that stop you labeling what you build as a release if it didn't come from a clean and tagged state). >> I've got myself GnuPG. I think I could dry run the whole process, fixing my broken scripts as I go, as I did with the rest.I think the outcome would now have to be tagged v2.7.2b2 to avoid confusion. A parallel possibility is to post v2.7.2b1 somewhere and just link it from the website. >> > A v2.2.2b2 sounds perfectly fine to me if it turns out that I'm unable > to get the current version working, but I do want to give it another > try. Once I figure out what is needed I'll make sure to let everyone > know what the fix was. Yes please! We have to encapsulate all this magic in the scripts or the instructions. > -Frank Jeff |
|
From: Jeff A. <ja...@fa...> - 2019-10-26 12:56:07
|
Hi Frank: I have some time now to try again, and the access to run the whole process for myself (thanks). I'll dry run quietly with 2.7.2b1 and aim for an official b2, enlisting your help to make it known. I fixed the gpg thing (see https://bugs.jython.org/issue2814) although I quite like the look of https://central.sonatype.org/pages/apache-ant.html#signing-and-deployments-using-the-maven-ant-tasks as an alternative to exec-ing gpg. I think the requirements have tightened since the previous release so signing will not be the only blocker. I'm sure I can figure it out, but did you learn anything in your staging attempts that I can get the easy way? Jeff Jeff Allen On 13/10/2019 09:06, Jeff Allen wrote: > On 13/10/2019 05:40, fwi...@gm... wrote: >> On Sat, Oct 12, 2019 at 2:19 PM Jeff Allen<ja...@fa...> wrote: >>> Hi Frank: >>> >>> Evidently this is keeping you up late. If it hasn't suddenly come right for you, I think I should take this back: seems we're not ready. >> No worries! Since I'm the only one that has released Jython in the >> last decade or so, I'm sure there is something peculiar in my >> environment that is interacting poorly with the current changes. I bet >> if I grab an old version of build.xml I may be able to get the process >> to work. I don't have time tonight but I'm sure I can give it a try >> tomorrow. > > Ok, well let us know when you've had enough :). Meanwhile I'll learn > what I can about the process. The video > <https://youtu.be/dXR4pJ_zS-0?t=300> suggests that when it fails > there's quite a lot of evidence where the bundle contents are defective. > > An old ./maven/build.xml is worth trying. I'm confident the current > main ./build.xml delivers the right things. The safeguards are all in > the main one (that stop you labeling what you build as a release if it > didn't come from a clean and tagged state). > >>> I've got myself GnuPG. I think I could dry run the whole process, fixing my broken scripts as I go, as I did with the rest.I think the outcome would now have to be tagged v2.7.2b2 to avoid confusion. A parallel possibility is to post v2.7.2b1 somewhere and just link it from the website. >>> >> A v2.2.2b2 sounds perfectly fine to me if it turns out that I'm unable >> to get the current version working, but I do want to give it another >> try. Once I figure out what is needed I'll make sure to let everyone >> know what the fix was. > Yes please! We have to encapsulate all this magic in the scripts or > the instructions. >> -Frank > Jeff |
|
From: Jeff A. <ja...@fa...> - 2019-10-30 21:07:33
|
Ok, I think I have succeeded in getting acceptable bundles for the -slim and -installer versions to stage, although not by a reproducible process. https://oss.sonatype.org/content/repositories/orgpython-1079/org/python/jython-installer/2.7.2b1/ https://oss.sonatype.org/content/repositories/orgpython-1078/org/python/jython-slim/2.7.2b1/ If I've understood correctly, this is one button-push away from publishing an actual release. However, I'll go through it all cleanly (that is, committed, tagged and using scripts) and call the result 2.7.2b2. Jeff Allen On 26/10/2019 13:55, Jeff Allen wrote: > > Hi Frank: > > I have some time now to try again, and the access to run the whole > process for myself (thanks). I'll dry run quietly with 2.7.2b1 and aim > for an official b2, enlisting your help to make it known. > > I fixed the gpg thing (see https://bugs.jython.org/issue2814) although > I quite like the look of > https://central.sonatype.org/pages/apache-ant.html#signing-and-deployments-using-the-maven-ant-tasks > as an alternative to exec-ing gpg. I think the requirements have > tightened since the previous release so signing will not be the only > blocker. > > I'm sure I can figure it out, but did you learn anything in your > staging attempts that I can get the easy way? > > Jeff > > Jeff Allen > On 13/10/2019 09:06, Jeff Allen wrote: >> On 13/10/2019 05:40, fwi...@gm... wrote: >>> On Sat, Oct 12, 2019 at 2:19 PM Jeff Allen<ja...@fa...> wrote: >>>> Hi Frank: >>>> >>>> Evidently this is keeping you up late. If it hasn't suddenly come right for you, I think I should take this back: seems we're not ready. >>> No worries! Since I'm the only one that has released Jython in the >>> last decade or so, I'm sure there is something peculiar in my >>> environment that is interacting poorly with the current changes. I bet >>> if I grab an old version of build.xml I may be able to get the process >>> to work. I don't have time tonight but I'm sure I can give it a try >>> tomorrow. >> >> Ok, well let us know when you've had enough :). Meanwhile I'll learn >> what I can about the process. The video >> <https://youtu.be/dXR4pJ_zS-0?t=300> suggests that when it fails >> there's quite a lot of evidence where the bundle contents are defective. >> >> An old ./maven/build.xml is worth trying. I'm confident the current >> main ./build.xml delivers the right things. The safeguards are all in >> the main one (that stop you labeling what you build as a release if >> it didn't come from a clean and tagged state). >> >>>> I've got myself GnuPG. I think I could dry run the whole process, fixing my broken scripts as I go, as I did with the rest.I think the outcome would now have to be tagged v2.7.2b2 to avoid confusion. A parallel possibility is to post v2.7.2b1 somewhere and just link it from the website. >>>> >>> A v2.2.2b2 sounds perfectly fine to me if it turns out that I'm unable >>> to get the current version working, but I do want to give it another >>> try. Once I figure out what is needed I'll make sure to let everyone >>> know what the fix was. >> Yes please! We have to encapsulate all this magic in the scripts or >> the instructions. >>> -Frank >> Jeff > > > _______________________________________________ > Jython-dev mailing list > Jyt...@li... > https://lists.sourceforge.net/lists/listinfo/jython-dev |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X