[reporter="edan", created="Wed, 8 Mar 2006 14:44:42 -0500 (GMT-05:00)"]
The following files were writable after an 'ant clean', 'ant cleanCatalog' in
farrago and an 'ant clean' in saffron. Farrago unit tests were failing. Deleting
these files, except for demo/hsqldb/scott.* in saffron, and doing an initBuild
caused the unit tests to function. (Note: the above clean commands and initBuild
were not enough to fix the problem).
There should be a target for all projects that can return the project to an
actual clean state. I don't know if these files should be added to the 'clean'
target, the 'cleanCatalog' target, or a new target.
------- Comment #2 From John Sichi 2004-06-17 10:03 [reply] -------
Downgrading this bug big time. It's very unlikely that any of these files
caused your problem. They hang around all the time and don't cause any trouble.
The one shady character is FarragoTrace.log.1.lck, which indicates that you may
have had an extra Java process hanging around screwing things up. Something
else you did in between might have killed that process. But since I don't have
any other diagnostics, it's hard to say.
------- Comment #3 From John Kalucki 2004-06-17 10:26 [reply] -------
Since these files cannot yet be ruled out as the cause of the instability:
1) My environment hasn't changed
2) I've had this same problem since 6/8.
3) I've run sqllineEngine successfully for ad-hoc testing in the mean time. (Not
likely to be a stale java process?)
4) Things started working, mysteriously, sometime fairly soon after removing
these files
Seems like we should at least eliminate these files as a potential problem.
This seems like a P1 to me.
------- Comment #4 From John Sichi 2004-06-17 10:46 [reply] -------
If you can pinpoint a particular leftover file that can consistently reproduces
a specific problem, we can reprioritize based on that. I don't see anything
suspicious in this list except the trace file I mentioned. The txn?.dat are
leftover Fennel transaction logs from crashes that aren't getting cleaned up
currently, but they shouldn't cause trouble.
If I had diagnostics from the original problem, it would be easier to
speculate.. Without even that, there's not much to go on.
[reporter="edan", created="Wed, 8 Mar 2006 14:44:42 -0500 (GMT-05:00)"] The following files were writable after an 'ant clean', 'ant cleanCatalog' in
farrago and an 'ant clean' in saffron. Farrago unit tests were failing. Deleting
these files, except for demo/hsqldb/scott.* in saffron, and doing an initBuild
caused the unit tests to function. (Note: the above clean commands and initBuild
were not enough to fix the problem).
There should be a target for all projects that can return the project to an
actual clean state. I don't know if these files should be added to the 'clean'
target, the 'cleanCatalog' target, or a new target.
farrago/catalog/xmi/NewAspenModelUML.xmi
farrago/catalog/xmi/FarragoConf.xmi
farrago/catalog/backup/FarragoCatalog.btd
farrago/catalog/backup/FarragoCatalog.btx
farrago/catalog/backup/db.dat
farrago/unitsql/ddl/mdr.btx
farrago/unitsql/ddl/mdr.btd
farrago/initBuild.properties
farrago/classpath.gen
farrago/trace/FarragoTrace.properties
farrago/trace/FarragoTrace.log.1.lck
farrago/trace/FarragoTrace.properties~
farrago/trace/FarragoTrace.log.1
farrago/txn1.dat
farrago/txn2.dat
farrago/testgen/FarragoSorterTest/data/testInMemorySparse.csv
farrago/testgen/FarragoSorterTest/data/testInMemoryDups.csv
farrago/testgen/FarragoSorterTest/data/testInMemoryDegenerate.csv
farrago/testgen/FarragoSorterTest/data/testInMemoryPresortedSparse.csv
farrago/testgen/FarragoSorterTest/data/testInMemoryPresortedDups.csv
farrago/testgen/FarragoSorterTest/data/testInMemoryPresortedSparseReverse.csv
farrago/testgen/FarragoSorterTest/data/testInMemoryPresortedDupsReverse.csv
farrago/txn4.dat
saffron/src/net/sf/saffron/sql/parser/ParserTokenManager.java
saffron/src/net/sf/saffron/sql/parser/ParserConstants.java
Not deleted, as they check out from Perforce as writable:
saffron/demo/hsqldb/scott.script
saffron/demo/hsqldb/scott.properties
------- Comment #1 From John Kalucki 2004-06-17 07:15 [reply] -------
Was failing in:
[junit] Running net.sf.farrago.test.FarragoSqlTest
[junit] Tests run: 25, Failures: 0, Errors: 0, Time elapsed: 163.691 sec
Added CCs.
------- Comment #2 From John Sichi 2004-06-17 10:03 [reply] -------
Downgrading this bug big time. It's very unlikely that any of these files
caused your problem. They hang around all the time and don't cause any trouble.
The one shady character is FarragoTrace.log.1.lck, which indicates that you may
have had an extra Java process hanging around screwing things up. Something
else you did in between might have killed that process. But since I don't have
any other diagnostics, it's hard to say.
------- Comment #3 From John Kalucki 2004-06-17 10:26 [reply] -------
Since these files cannot yet be ruled out as the cause of the instability:
1) My environment hasn't changed
2) I've had this same problem since 6/8.
3) I've run sqllineEngine successfully for ad-hoc testing in the mean time. (Not
likely to be a stale java process?)
4) Things started working, mysteriously, sometime fairly soon after removing
these files
Seems like we should at least eliminate these files as a potential problem.
This seems like a P1 to me.
------- Comment #4 From John Sichi 2004-06-17 10:46 [reply] -------
If you can pinpoint a particular leftover file that can consistently reproduces
a specific problem, we can reprioritize based on that. I don't see anything
suspicious in this list except the trace file I mentioned. The txn?.dat are
leftover Fennel transaction logs from crashes that aren't getting cleaned up
currently, but they shouldn't cause trouble.
If I had diagnostics from the original problem, it would be easier to
speculate.. Without even that, there's not much to go on.