A plan is better than no plan, so here is my suggestion:
The Pike 7.3 CVS is forked into a new development version, 7.5, next saturday (14/9). At the same time the 7.3 tree is renamed to 7.4. All API/language design changes and the like must be completed before then. Only bug fixes and release related changes will be accepted after the CVS shuffle.
Bugs: 3022 BMP fails on 64bit machines (was this a WONTFIX? If so the test should be rewritten) FreeBSD sometimes fails to build post_modules and to run popen. Some machines has problems running Crypto.sha correctly. MS Windows fails spectaculary with dynamic loading. Turn it off?
Design issues: There must be a semicolon added after implicit lambdas.
What's missing/wrong?
I have debugged the MS Windows failure a bit, and it's probably related to a broken dynamic Gmp lib. I think I can have that fixed by next saturday.
/ Peter Bortas
Previous text:
2002-09-08 16:01: Subject: Plan
A plan is better than no plan, so here is my suggestion:
The Pike 7.3 CVS is forked into a new development version, 7.5, next saturday (14/9). At the same time the 7.3 tree is renamed to 7.4. All API/language design changes and the like must be completed before then. Only bug fixes and release related changes will be accepted after the CVS shuffle.
Bugs: 3022 BMP fails on 64bit machines (was this a WONTFIX? If so the test should be rewritten) FreeBSD sometimes fails to build post_modules and to run popen. Some machines has problems running Crypto.sha correctly. MS Windows fails spectaculary with dynamic loading. Turn it off?
Design issues: There must be a semicolon added after implicit lambdas.
What's missing/wrong?
/ Martin Nilsson (Fake Build Master)
If someone can kidnap a nonbroken Gmp from RIS I can have it fixed this afternoon. :)
/ Peter Bortas
Previous text:
2002-09-08 16:05: Subject: Plan
I have debugged the MS Windows failure a bit, and it's probably related to a broken dynamic Gmp lib. I think I can have that fixed by next saturday.
/ Peter Bortas
Grubba, has the new program encode/decode-value stuff in 7.2 been ported to 7.3?
/ Martin Stjernholm, Roxen IS
Previous text:
2002-09-08 16:01: Subject: Plan
A plan is better than no plan, so here is my suggestion:
The Pike 7.3 CVS is forked into a new development version, 7.5, next saturday (14/9). At the same time the 7.3 tree is renamed to 7.4. All API/language design changes and the like must be completed before then. Only bug fixes and release related changes will be accepted after the CVS shuffle.
Bugs: 3022 BMP fails on 64bit machines (was this a WONTFIX? If so the test should be rewritten) FreeBSD sometimes fails to build post_modules and to run popen. Some machines has problems running Crypto.sha correctly. MS Windows fails spectaculary with dynamic loading. Turn it off?
Design issues: There must be a semicolon added after implicit lambdas.
What's missing/wrong?
/ Martin Nilsson (Fake Build Master)
Not as such, but it complicates the port a bit since it's the same code. What the encode/decode fixes in 7.2 do is to abstract the dumping of the symbol tables in programs in a platform independent way. Together with the non-native byte code they make program dumps architecture independent. They also improves the error checking when dumped programs are loaded and e.g. the programs they inherit have changed, and they are a step towards making dumped programs less dependent on the exact storage structure of inherited programs.
/ Martin Stjernholm, Roxen IS
Previous text:
2002-09-09 16:37: Subject: Plan
Is the native machine code stuff an issue here?
/ Martin Nilsson (Fake Build Master)
I should make "local::" on variables an error. C.f. [bug 1256].
/ Martin Stjernholm, Roxen IS
Previous text:
2002-09-08 16:01: Subject: Plan
A plan is better than no plan, so here is my suggestion:
The Pike 7.3 CVS is forked into a new development version, 7.5, next saturday (14/9). At the same time the 7.3 tree is renamed to 7.4. All API/language design changes and the like must be completed before then. Only bug fixes and release related changes will be accepted after the CVS shuffle.
Bugs: 3022 BMP fails on 64bit machines (was this a WONTFIX? If so the test should be rewritten) FreeBSD sometimes fails to build post_modules and to run popen. Some machines has problems running Crypto.sha correctly. MS Windows fails spectaculary with dynamic loading. Turn it off?
Design issues: There must be a semicolon added after implicit lambdas.
What's missing/wrong?
/ Martin Nilsson (Fake Build Master)
Bugs: --with-machine-code function calls on AIX/POWER.
Should we put 7.5 directly into Subversion, or make a CVS repository first and move it when CL etc is ready for it?
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
Previous text:
2002-09-08 16:01: Subject: Plan
A plan is better than no plan, so here is my suggestion:
The Pike 7.3 CVS is forked into a new development version, 7.5, next saturday (14/9). At the same time the 7.3 tree is renamed to 7.4. All API/language design changes and the like must be completed before then. Only bug fixes and release related changes will be accepted after the CVS shuffle.
Bugs: 3022 BMP fails on 64bit machines (was this a WONTFIX? If so the test should be rewritten) FreeBSD sometimes fails to build post_modules and to run popen. Some machines has problems running Crypto.sha correctly. MS Windows fails spectaculary with dynamic loading. Turn it off?
Design issues: There must be a semicolon added after implicit lambdas.
What's missing/wrong?
/ Martin Nilsson (Fake Build Master)
I prefer if we wait with moving to subversion until CL, Xenofarm (and more?) is svn-enabled. It is not an answer to your question, but I don't think we'll manage to be ready in just one more week. I prefer focusing on Pike.
/ Martin Nilsson (Fake Build Master)
Previous text:
2002-09-08 16:50: Subject: Plan
Bugs: --with-machine-code function calls on AIX/POWER.
Should we put 7.5 directly into Subversion, or make a CVS repository first and move it when CL etc is ready for it?
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
I also think we should focus 7.4 first and put it in subversion later. Regarding the cvs browser, I won't get it svn-ready until the release. A pike with svn bindings @ pelix would probably get the code librarian svn support there quicker, though. :-) Especially after it has the fs API. IIRC the RA API could be fun too, but it's possible I would have to run the indexer locally at all times anyway, unless the svn guys thought about that use case.
/ Johan Sundström (nu även i Gottfridsberg!)
Previous text:
2002-09-08 16:53: Subject: Plan
I prefer if we wait with moving to subversion until CL, Xenofarm (and more?) is svn-enabled. It is not an answer to your question, but I don't think we'll manage to be ready in just one more week. I prefer focusing on Pike.
/ Martin Nilsson (Fake Build Master)
pike-devel@lists.lysator.liu.se
-
Henrik Grubbstr�m (Lysator) ＠ Pike (-) developers forum -
Johan Sundstr�m (nu �ven i Gottfridsberg!) ＠ Pike (-) developers forum
-
Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum
-
Martin Nilsson (Fake Build Master) ＠ Pike (-) developers forum
-
Martin Stjernholm, Roxen IS ＠ Pike developers forum
-
Peter Bortas ＠ Pike developers forum