Concatenating CHAR(N) and VARCHAR strings
Hello, What happens when you concatenate a CHAR(N) and a VARCHAR value? Calcite seems to insert no casts before concatenation, so the trailing spaces for the CHAR(N) value are preserved. However, Postgres has the following test: https://github.com/postgres/postgres/blob/5bcc7e6dc8cb6fb6ab16c118a1e710fc85 096f76/src/test/regress/expected/strings.out#L1802C1-L1805C21 SELECT char(20) 'characters' || ' and text' AS "Concat char to unknown type"; Concat char to unknown type - characters and text The same test (adapted for the Babel syntax) evaluates statically in Calcite to: "characters and text" It looks like Postgres implicitly casts the CHAR(20) value to TEXT before concatenation. I have stared at tables with type conversions in various databases for a long time, but I could not figure out what the result is supposed to be. Thank you, Mihai
[jira] [Created] (CALCITE-5811) Error messages produced for constant out-of-bounds arguments are confusing
Mihai Budiu created CALCITE-5811:
Summary: Error messages produced for constant out-of-bounds
arguments are confusing
Key: CALCITE-5811
URL: https://issues.apache.org/jira/browse/CALCITE-5811
Project: Calcite
Issue Type: Bug
Affects Versions: 1.34.0
Reporter: Mihai Budiu
Supplying arguments that are out-of-bounds for functions can produce
uninformative messages. Consider this test case that can be added to
SqlOperatorTest.checkSubstringFunction():
{code:java}
f.checkString("substring('abc' from 2 for 2147483650)",
"bc", "VARCHAR(3) NOT NULL");
{code}
The signature of the substring function requires int arguments. The constant
2147483650 is out of bounds for an integer. This causes the test to fail with
the following exception:
{code:java}
Error while executing SQL "values (substring('abc' from 2 for 2147483650))":
Unable to implement EnumerableCalc(expr#0=[{inputs}], expr#1=['abc'],
expr#2=[2], expr#3=[2147483650:BIGINT], expr#4=[SUBSTRING($t1, $t2, $t3)],
EXPR$0=[$t4]): rowcount = 1.0, cumulative cost = {2.0 rows, 7.0 cpu, 0.0 io},
id = 153424
EnumerableValues(tuples=[[{ 0 }]]): rowcount = 1.0, cumulative cost = {1.0
rows, 1.0 cpu, 0.0 io}, id = 153408
{code}
I suspect this happens because of a combination of features:
- type inference does not reject the call to substring, although the inferred
type of the argument is probably bigint
- an attempt to evaluate the constant function fails because of some overflow
- the detailed exception about the overflow is lost when handling the error
I suspect this is a deeper problem which may affect all functions, and it is
really about type inference and implicit cast insertion.
I can imagine two possible implementation strategies:
- Postgres rejects such a call because the substring function cannot take a
bigint argument
- The compiler could add an implicit cast from bigint to int, which should at
least give a warning because the value is too large to fit in an integer
Both these solutions would avoid a crash in the expression evaluation. I
personally do not yet understand well enough the type inference mechanisms in
Calcite to propose a solution.
To make matters worse, some SQL dialects have substring functions with bigint
arguments. In this case there should be two different substring functions, with
different signatures.
This issue surfaced during a discussion for
[https://github.com/apache/calcite/pull/3286,] and is related to
https://issues.apache.org/jira/browse/CALCITE-5810.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
Re: Virtual key signing party
Signed key attached. -- Michael Mior mm...@apache.org On Tue, Jun 27, 2023 at 3:32 AM Ruben Q L wrote: > Hello, > > I'll try to attend too. > Best, > Ruben > > > pub rsa4096 2020-09-26 [SC] > > 464F A4A3 D7E4 2112 E4CB 68F2 DF92 5FEB A08B 032B > > uid[ absoluta ] Ruben Quesada Lopez > > sub rsa4096 2020-09-26 [E] > > > > > On Tue, Jun 27, 2023 at 8:29 AM Stamatis Zampetakis > wrote: > > > Great let's do one on June 29, 2023, 14:30 UTC+2 [1, 2]. For those who > > can't make it in this slot feel free to propose additional ones. > > > > Best, > > Stamatis > > > > [1] https://s.apache.org/cn5xq > > [2] https://meet.google.com/vwx-mxxz-ibk > > > > > > On Tue, Jun 27, 2023 at 4:22 AM xiong duan wrote: > > > > > > Hi Stamatis, > > > Thanks for organizing this party. I am in UTC+8. How about June 29, > > > 2023, 20:30 (UTC+8) So the UTC+2 time should be June 29, 2023, 14:30? > > > Here is my fingerprint: > > > > > > pub rsa4096 2023-06-27 [SC] > > > > > > EC78 877D 98B8 CDD4 9613 3DF2 B989 C696 D102 A552 > > > > > > uid [ultimate] Xiong Duan (CODE SIGNING KEY) < > xi...@apache.org > > > > > > > > > sub rsa4096 2023-06-27 [E] > > > > > > Michael Mior 于2023年6月27日周二 02:19写道： > > > > > > > I'll attend if possible. My key needs to be updated. (Not > compromised, > > but > > > > I lost access to my previous signing key.) > > > > > > > > pub rsa4096 2022-06-17 [SC] [expires: 2038-06-13] > > > > EAC5 89C4 44F4 68FE 7E11 3D2D D8D7 2C13 CF2D 8DDB > > > > uid [ultimate] Michael Mior > > > > uid [ultimate] Michael Mior > > > > uid [ultimate] Michael Mior > > > > sub rsa4096 2022-06-17 [E] [expires: 2038-06-13] > > > > > > > > -- > > > > Michael Mior > > > > mm...@apache.org > > > > > > > > > > > > On Mon, Jun 26, 2023 at 3:27 AM Stamatis Zampetakis < > zabe...@gmail.com > > > > > > > wrote: > > > > > > > > > Hello, > > > > > > > > > > The last virtual key signing party [1] was about a year ago [2]. In > > > > > the upcoming releases, we have some people that are serving as > > release > > > > > managers for the first time, thus it may be a good time to hold > > > > > another party. > > > > > > > > > > Since Duan Xiong is the RM for the next release, I would suggest > > > > > holding a meeting in a timezone that is convenient for him. > > > > > > > > > > @Duan: Can you please propose a time slot that works for you? > > > > > > > > > > I am in UTC+2 and will try to attend if the slot doesn't fall in > the > > > > > middle of the night (23:00 - 06:00). > > > > > > > > > > For those that would like to attend, please reply to this thread > with > > > > > your public > > > > > keys' fingerprint (gpg --fingerprint) before the meeting. > > > > > > > > > > pub rsa4096 2019-03-15 [SC] [expires: 2027-03-15] > > > > > 0474 9577 FD93 4674 B9CD 45C5 D77C 3383 F192 7570 > > > > > uid [ultimate] Stamatis Zampetakis > > > > > uid [ultimate] Stamatis Zampetakis > > > > > sub rsa4096 2019-03-15 [E] [expires: 2027-03-15] > > > > > > > > > > To verify your identity, please bring with you a government issued > ID > > > > > (preferably passport). > > > > > > > > > > As a reminder of the procedure, have a look at the notes [1] taken > by > > > > > Francis > > > > > during a previous party! > > > > > > > > > > Anybody can participate (not need to be a committer to the > project). > > > > > > > > > > Best, > > > > > Stamatis > > > > > > > > > > [1] > > > > > > > > > > > > > > > > > http://www.cryptnet.net/fdp/crypto/keysigning_party/en/keysigning_party.html > > > > > [2] > https://lists.apache.org/thread/4lvch7sfwszbylpzjt11jc9tokm3d0l4 > > > > > [3] https://gist.github.com/F21/b0e8c62c49dfab267ff1d0c6af39ab84 > > > > > > > > > > > >
Re: Virtual key signing party
Signed key attached. -- Michael Mior mm...@apache.org On Mon, Jun 26, 2023 at 10:22 PM xiong duan wrote: > Hi Stamatis, > Thanks for organizing this party. I am in UTC+8. How about June 29, > 2023, 20:30 (UTC+8) So the UTC+2 time should be June 29, 2023, 14:30? > Here is my fingerprint: > > pub rsa4096 2023-06-27 [SC] > > EC78 877D 98B8 CDD4 9613 3DF2 B989 C696 D102 A552 > > uid [ultimate] Xiong Duan (CODE SIGNING KEY) > > sub rsa4096 2023-06-27 [E] > > Michael Mior 于2023年6月27日周二 02:19写道： > > > I'll attend if possible. My key needs to be updated. (Not compromised, > but > > I lost access to my previous signing key.) > > > > pub rsa4096 2022-06-17 [SC] [expires: 2038-06-13] > > EAC5 89C4 44F4 68FE 7E11 3D2D D8D7 2C13 CF2D 8DDB > > uid [ultimate] Michael Mior > > uid [ultimate] Michael Mior > > uid [ultimate] Michael Mior > > sub rsa4096 2022-06-17 [E] [expires: 2038-06-13] > > > > -- > > Michael Mior > > mm...@apache.org > > > > > > On Mon, Jun 26, 2023 at 3:27 AM Stamatis Zampetakis > > wrote: > > > > > Hello, > > > > > > The last virtual key signing party [1] was about a year ago [2]. In > > > the upcoming releases, we have some people that are serving as release > > > managers for the first time, thus it may be a good time to hold > > > another party. > > > > > > Since Duan Xiong is the RM for the next release, I would suggest > > > holding a meeting in a timezone that is convenient for him. > > > > > > @Duan: Can you please propose a time slot that works for you? > > > > > > I am in UTC+2 and will try to attend if the slot doesn't fall in the > > > middle of the night (23:00 - 06:00). > > > > > > For those that would like to attend, please reply to this thread with > > > your public > > > keys' fingerprint (gpg --fingerprint) before the meeting. > > > > > > pub rsa4096 2019-03-15 [SC] [expires: 2027-03-15] > > > 0474 9577 FD93 4674 B9CD 45C5 D77C 3383 F192 7570 > > > uid [ultimate] Stamatis Zampetakis > > > uid [ultimate] Stamatis Zampetakis > > > sub rsa4096 2019-03-15 [E] [expires: 2027-03-15] > > > > > > To verify your identity, please bring with you a government issued ID > > > (preferably passport). > > > > > > As a reminder of the procedure, have a look at the notes [1] taken by > > > Francis > > > during a previous party! > > > > > > Anybody can participate (not need to be a committer to the project). > > > > > > Best, > > > Stamatis > > > > > > [1] > > > > > > > > > http://www.cryptnet.net/fdp/crypto/keysigning_party/en/keysigning_party.html > > > [2] https://lists.apache.org/thread/4lvch7sfwszbylpzjt11jc9tokm3d0l4 > > > [3] https://gist.github.com/F21/b0e8c62c49dfab267ff1d0c6af39ab84 > > > > > >
