| 1 | * easye reports on work on abcl-1.1.0 for 20120805 |
|---|
| 2 | |
|---|
| 3 | Not much in actual commits from me at this point, just some basic |
|---|
| 4 | tightening of the logic in the build process and README refreshes. |
|---|
| 5 | |
|---|
| 6 | My main energy, such as it was, involved working on setting up a |
|---|
| 7 | quasi-continuous built/test environment for ABCL on various OS/JVM |
|---|
| 8 | combinations. Noticibly deficient is access to a reasonably "hands |
|---|
| 9 | off" MSFT Windows environment, but with suitable virtualization, I |
|---|
| 10 | could do something. The open question: what would be the MSFT Windows |
|---|
| 11 | equivalent to /etc/rc.local? Is this even possible on a stock MSFT |
|---|
| 12 | install? Tentative conclusion: just run the Windows tests manually |
|---|
| 13 | as the effort involved here exceeds probably return on investment. |
|---|
| 14 | |
|---|
| 15 | ** bisecting where changes take place |
|---|
| 16 | |
|---|
| 17 | If someone is able to create a test Bourne shell script that invokes |
|---|
| 18 | abcl in a failing manner, I can submit this to my build farm to find |
|---|
| 19 | out which commit actually breaks the trunk. Please feel free to bug |
|---|
| 20 | me on #abcl if this service could help you |
|---|
| 21 | |
|---|
| 22 | ** ABCL-CONTRIB JNA fails to load on Solaris |
|---|
| 23 | |
|---|
| 24 | My build farm is hosted on solaris-x86_64 which seems to have caught a |
|---|
| 25 | corner case in the JAR-PATHNAME code, looking like an unexpected |
|---|
| 26 | DEVICE for a jar being bound to a "lingering" special. Does not occur |
|---|
| 27 | on osx-x86_64, which is rather puzzling. Need to recheck my |
|---|
| 28 | environments to ensure that I don't have an inadvertent patch applied |
|---|
| 29 | somewhere. |
|---|
| 30 | |
|---|
| 31 | ** Next Week |
|---|
| 32 | |
|---|
| 33 | *** Start testing whitespace in Pathnames |
|---|
| 34 | |
|---|
| 35 | ABCL-TEST-LISP tests are failing that used to work, so the starting |
|---|
| 36 | point is to start analyzing where they fail. Fortunately, I can now |
|---|
| 37 | bisect the offending changes in a batch mode at this point. |
|---|
| 38 | |
|---|
| 39 | *** ABCL-ASDF testing |
|---|
| 40 | |
|---|
| 41 | After figuring out the ABCL-CONTRIB JNA bug, more testing of the Maven |
|---|
| 42 | Aether interface is in order. Goal: load CFFI without user |
|---|
| 43 | configuration in more places. |
|---|
| 44 | |
|---|
| 45 | |
|---|
| 46 | |
|---|
| 47 | |
|---|
| 48 | |
|---|
| 49 | |
|---|
| 50 | |
|---|
