Greetings, Warren Young!
>>> None of the SQLite core developers have responded to my charge that this
>>> looks like a bug in SQLite. It shouldn't be generating temporary file
>>> names with backslashes in them for Cygwin builds...
>>
>> There's no reason to ever use backslashes in paths, ever.
Warren Young writes:
> I'm partially waiting to push this test release to curr on your
> behalf. I expect that with CYGWIN_SQLITE_LOCKING=posix, you will be
> able to stop using your local version of sqlite3.
I've actually built a 3.7.17 version that has been sitting around
untested for a while do
On 6/12/2013 12:13, Achim Gratz wrote:
Warren Young writes:
If true, that's yet another reason to switch to 3.7.17: I removed the
hack since we're building as SQLITE_OS_UNIX.
I can attest that this solves the problem with temp files, as expected.
Thanks for testing!
I'm sorry, but I haven'
Warren Young writes:
> If true, that's yet another reason to switch to 3.7.17: I removed the
> hack since we're building as SQLITE_OS_UNIX.
I can attest that this solves the problem with temp files, as expected.
I'm sorry, but I haven't been able to do much else with that test
version.
Regards,
On 6/11/2013 18:19, Andrey Repin wrote:
None of the SQLite core developers have responded to my charge that this
looks like a bug in SQLite. It shouldn't be generating temporary file
names with backslashes in them for Cygwin builds...
There's no reason to ever use backslashes in paths, ever.
Greetings, Warren Young!
> On 6/9/2013 19:26, Daniel Colascione wrote:
>> which I haven't been able to test
> You should. One of the changes is to prefer creating temporary tables
> in memory instead of on disk, which should bypass the problem.
>> "/var/tmp/etilqs_z28HceqmzVr3ZO1\\etilqs_rnPCu
On 6/10/2013 14:34, Warren Young wrote:
There's only 140 kSLOC in SQLite. Easy.
...of which about 6 kSLOC changed between the releases you're comparing.
I'm not going to plow through a diff that big any time soon.
--
Problem reports: http://cygwin.com/problems.html
FAQ:
On 6/10/2013 14:21, Daniel Colascione wrote:
On 6/10/2013 12:21 PM, Warren Young wrote:
On 6/9/2013 19:26, Daniel Colascione wrote:
which I haven't been able to test
You should. One of the changes is to prefer creating temporary tables in memory
instead of on disk, which should bypass the pr
On 6/10/2013 12:21 PM, Warren Young wrote:
> On 6/9/2013 19:26, Daniel Colascione wrote:
>> which I haven't been able to test
>
> You should. One of the changes is to prefer creating temporary tables in
> memory
> instead of on disk, which should bypass the problem.
This change makes me nervous
On 6/9/2013 19:26, Daniel Colascione wrote:
which I haven't been able to test
You should. One of the changes is to prefer creating temporary tables
in memory instead of on disk, which should bypass the problem.
"/var/tmp/etilqs_z28HceqmzVr3ZO1\\etilqs_rnPCuceSOgjfeTd".
This bug is alread
The mandatory locking work (which I haven't been able to test) aside, temporary
table creation is broken with SQLite 3.7.16.2-1. 'CREATE TEMP TABLE foo (bar
INT)' fails. cygwinGetTempname, called from getTempname, returns the correct
temporary directory --- "/var/tmp/etilqs_z28HceqmzVr3ZO1" in my c
11 matches
Mail list logo
