[EMAIL PROTECTED] wrote:
My guess is that your virus scanning software is opening
the test.db file as sson as SQLite closes it, in order to
check it for new viruses. The test script tries to delete
the file right after closing it, but it can't because the
virus scanner has it open.
I don't have a virus scanner on this PC. It sits behind a virus scanning
firewall, and they trust us enough not to put in virus laden disks. I
don't have any indexing software either. I did have an explorer window
open displaying that directory, but I get the same result with that
window closed.
I just skipped the exclusive2 tests by editing the exclude list in the
quick test file. All other test run to completion. I got one error due
to differences in floating point exponent formating between TCL on
windows and linux (at least that is my suspicion).
alter-7.1...
Expected: [text 1 integer -2 text 5.4e-8 real 5.4e-08]
Got: [text 1 integer -2 text 5.4e-8 real 5.4e-008]
I just can't see why it can't delete that file.
I modified the exclusive2.test file and wrapped the file delete commands
in catch blocks so they wouldn't halt the test, and then changed the
name of the database file used for the 3.x tests from test.db to
test3.db. I also added a pair of file delete commands in catch blocks to
remove the test3.db file. Now when I run this test set I get no failures.
$ ./testfixture.exe ../sqlite/test/exclusive2.test
exclusive2-1.0... Ok
exclusive2-1.1... Ok
exclusive2-1.2... Ok
exclusive2-1.3... Ok
exclusive2-1.4... Ok
exclusive2-1.5... Ok
exclusive2-1.6... Ok
exclusive2-1.7... Ok
exclusive2-1.9... Ok
exclusive2-1.10... Ok
exclusive2-1.11... Ok
exclusive2-2.1... Ok
exclusive2-2.2... Ok
exclusive2-2.3... Ok
exclusive2-2.4... Ok
exclusive2-2.5... Ok
exclusive2-2.6... Ok
exclusive2-2.7... Ok
exclusive2-2.8... Ok
exclusive2-3.0... Ok
exclusive2-3.1... Ok
exclusive2-3.2... Ok
exclusive2-3.3... Ok
exclusive2-3.4... Ok
exclusive2-3.5... Ok
exclusive2-3.6... Ok
Thread-specific data deallocated properly
0 errors out of 27 tests
Failures on these tests:
But I can see that neither the test.db file or the new test3.db were
deleted.
The only thing I can see that is common to these files is that they are
both passed to the pagerChangeCounter command which is only used in this
test.
proc pagerChangeCounter {filename {new ""}} {
set fd [open $filename RDWR]
fconfigure $fd -translation binary -encoding binary
if {$new ne ""} {
seek $fd 24
set a [expr {($new&0xFF000000)>>24}]
set b [expr {($new&0x00FF0000)>>16}]
set c [expr {($new&0x0000FF00)>>8}]
set d [expr {($new&0x000000FF)}]
puts -nonewline $fd [binary format cccc $a $b $c $d]
flush $fd
}
seek $fd 24
foreach {a b c d} [list 0 0 0 0] {}
binary scan [read $fd 4] cccc a b c d
set ret [expr ($a&0x000000FF)<<24]
incr ret [expr ($b&0x000000FF)<<16]
incr ret [expr ($c&0x000000FF)<<8]
incr ret [expr ($d&0x000000FF)<<0]
close $fd
return $ret
}
I don't know enough TCL to know if there is a subtle bug here or not. I
suspect that perhaps the file isn't really being closed until the script
exits.Does this seem possible?
Dennis Cote
-----------------------------------------------------------------------------
To unsubscribe, send email to [EMAIL PROTECTED]
-----------------------------------------------------------------------------
