":memory:" path does not seem to work on Windows anymore

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

":memory:" path does not seem to work on Windows anymore

Kirill Müller
Hi


In the R interface to SQLite [1], we observe that opening a database
with ":memory:" does not work anymore on Windows (both 32- and 64-bit
versions), possibly related to an upgrade from sqlite 3.8.6 to 3.11.1.
Using "file::memory:" works on 3.11.1, but not for 3.8.6 . The
sqlite3_open_v2() function returns SQLITE_CANTOPEN, and is called with
(":memory:", a valid pointer to a sqlite3*, SQLITE_RWC, NULL) according
to code inspection. Please advise.


Best regards

Kirill


[1] https://github.com/rstats-db/RSQLite
_______________________________________________
sqlite-users mailing list
[hidden email]
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
Reply | Threaded
Open this post in threaded view
|

Re: ":memory:" path does not seem to work on Windows anymore

Kirill Müller
I can't reproduce the issue on Windows with the current command-line
client, but it is real in our environment. How can I help you replicate
it? Thanks.


-Kirill

On 26.05.2016 14:57, Kirill Müller wrote:

> Hi
>
>
> In the R interface to SQLite [1], we observe that opening a database
> with ":memory:" does not work anymore on Windows (both 32- and 64-bit
> versions), possibly related to an upgrade from sqlite 3.8.6 to 3.11.1.
> Using "file::memory:" works on 3.11.1, but not for 3.8.6 . The
> sqlite3_open_v2() function returns SQLITE_CANTOPEN, and is called with
> (":memory:", a valid pointer to a sqlite3*, SQLITE_RWC, NULL)
> according to code inspection. Please advise.
>
>
> Best regards
>
> Kirill
>
>
> [1] https://github.com/rstats-db/RSQLite
> _______________________________________________
> sqlite-users mailing list
> [hidden email]
> http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users

_______________________________________________
sqlite-users mailing list
[hidden email]
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
Reply | Threaded
Open this post in threaded view
|

Re: ":memory:" path does not seem to work on Windows anymore

Stephan Beal-3
On Mon, May 30, 2016 at 8:35 AM, Kirill Müller <[hidden email]>
wrote:

> I can't reproduce the issue on Windows with the current command-line
> client, but it is real in our environment. How can I help you replicate it?
> Thanks.
>
>
> -Kirill
>
> On 26.05.2016 14:57, Kirill Müller wrote:
>
>> Hi
>>
>>
>> In the R interface to SQLite [1], we observe that opening a database with
>> ":memory:" does not work anymore on Windows (both 32- and 64-bit versions),
>
>
might it be a problem with the R wrapper? If it cannot be reproduced in the
cli client then the wrapper is the most likely culprit.

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Freedom is sloppy. But since tyranny's the only guaranteed byproduct of
those who insist on a perfect world, freedom will have to do." -- Bigby Wolf
_______________________________________________
sqlite-users mailing list
[hidden email]
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
Reply | Threaded
Open this post in threaded view
|

Re: ":memory:" path does not seem to work on Windows anymore

Kirill Müller
On 30.05.2016 09:57, Stephan Beal wrote:

> On Mon, May 30, 2016 at 8:35 AM, Kirill Müller <[hidden email]>
> wrote:
>
>> I can't reproduce the issue on Windows with the current command-line
>> client, but it is real in our environment. How can I help you replicate it?
>> Thanks.
>>
>>
>> -Kirill
>>
>> On 26.05.2016 14:57, Kirill Müller wrote:
>>
>>> Hi
>>>
>>>
>>> In the R interface to SQLite [1], we observe that opening a database with
>>> ":memory:" does not work anymore on Windows (both 32- and 64-bit versions),
>>
> might it be a problem with the R wrapper? If it cannot be reproduced in the
> cli client then the wrapper is the most likely culprit.
>
The R wrapper is rather thin at this point, perhaps the build
environment is triggering this. It used to work with an older version of
SQLite, thoujgh. I'd be happy to run a bisection against SQLite, but I'm
having trouble finding intermediate amalgamation sources, and I'm not
really familiar with Fossil. I'd appreciate any pointers on bisecting
SQLite. Thanks.


-Kirill
_______________________________________________
sqlite-users mailing list
[hidden email]
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
Reply | Threaded
Open this post in threaded view
|

Re: ":memory:" path does not seem to work on Windows anymore

Richard Hipp-3
On 5/30/16, Kirill Müller <[hidden email]> wrote:
>  I'd appreciate any pointers on bisecting
> SQLite. Thanks.

(1) Make sure you have tclsh 8.5 or later installed on your system, as
there are various TCL scripts that SQLite makefile needs to run in
order to build the amalgamation.

(2) Install Fossil.  You can either download the precompiled
stand-alone binary from https://www.fossil-scm.org/download.html and
put that binary somewhere on your $PATH, or you can compile the
stand-alone binary yourself, from sources.
(https://www.fossil-scm.org/fossil/doc/trunk/www/build.wiki)  Either
way, installing is just a matter of putting the "fossil" or
"fossil.exe" file on your $PATH or %PATH% and uninstalling is simply
deleting the binary.

(3) Clone the SQLite repo:  fossil clone http://www.sqlite.org/src sqlite.fossil

(4) Open a check-out on the SQLite repo:  fossil open sqlite.fossil

(5) Make sure you can build:  ./configure; make sqlite3.c

(6) Start the bisect:  "fossil bisect reset; fossil bisect good
version-3.11.0; fossil bisect bad trunk".  In place of
"version-3.11.0" and "trunk", substitute whatever tags and/or SHA-1
check-in hashes and/or ISO8601 timestamps are appropriate to define
the end-points of the bisection.

(7) Rebuild and test.  Then type "fossil bisect good|bad" as
appropriate.  "fossil bisect undo" is available if you make a mistake
and need to back up.

(8) When the bisect finishes (or at any time in between) you can do
"fossil bisect ui" to see a timeline graph of the bisection.
--
D. Richard Hipp
[hidden email]
_______________________________________________
sqlite-users mailing list
[hidden email]
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
Reply | Threaded
Open this post in threaded view
|

Re: ":memory:" path does not seem to work on Windows anymore

Kirill Müller
On 30.05.2016 17:09, Richard Hipp wrote:

> On 5/30/16, Kirill Müller <[hidden email]> wrote:
>>   I'd appreciate any pointers on bisecting
>> SQLite. Thanks.
> (1) Make sure you have tclsh 8.5 or later installed on your system, as
> there are various TCL scripts that SQLite makefile needs to run in
> order to build the amalgamation.
>
> (2) Install Fossil.  You can either download the precompiled
> stand-alone binary from https://www.fossil-scm.org/download.html and
> put that binary somewhere on your $PATH, or you can compile the
> stand-alone binary yourself, from sources.
> (https://www.fossil-scm.org/fossil/doc/trunk/www/build.wiki)  Either
> way, installing is just a matter of putting the "fossil" or
> "fossil.exe" file on your $PATH or %PATH% and uninstalling is simply
> deleting the binary.
>
> (3) Clone the SQLite repo:  fossil clone http://www.sqlite.org/src sqlite.fossil
>
> (4) Open a check-out on the SQLite repo:  fossil open sqlite.fossil
>
> (5) Make sure you can build:  ./configure; make sqlite3.c
>
> (6) Start the bisect:  "fossil bisect reset; fossil bisect good
> version-3.11.0; fossil bisect bad trunk".  In place of
> "version-3.11.0" and "trunk", substitute whatever tags and/or SHA-1
> check-in hashes and/or ISO8601 timestamps are appropriate to define
> the end-points of the bisection.
>
> (7) Rebuild and test.  Then type "fossil bisect good|bad" as
> appropriate.  "fossil bisect undo" is available if you make a mistake
> and need to back up.
>
> (8) When the bisect finishes (or at any time in between) you can do
> "fossil bisect ui" to see a timeline graph of the bisection.
Thanks for the very helpful detailed instructions. The original issue
probably comes from how we compile and link the SQLite files, this is
most likely not a regression.

I have noticed that fossil 1.33 is missing a "bisect skip" command which
would allow ignoring revisions that are "untestable" but cannot be
decided to be good or bad.


-Kirill
_______________________________________________
sqlite-users mailing list
[hidden email]
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users