Selects in Joins - sql

I have query
UPDATE THD
SET RepostFlag = 'Y'
,RunListNoRetroPolicyPrepay = ?
,RetroObject = ?
FROM TranHead AS THD
JOIN (
SELECT CustPolicyNo AS CustPolicyNo
,MIN(PremPeriod) AS PremPeriod
FROM TranHead
WHERE RepostFlag = 'Y'
AND PayoutTypeNo = ?
GROUP BY CustPolicyNo
) AS THDToBeReposted ON THD.CustPolicyNo = THDToBeReposted.CustPolicyNo
WHERE THD.RepostFlag = 'N'
AND THD.PremPeriod > THDToBeReposted.PremPeriod
fails in H2 with following message
Table "THD" not found;
I looked at http://www.h2database.com/html/grammar.html#table_expression to see if H2 supports selects in join. It appears it does. Maybe I am missing something when looking at the grammar, but it seems to me that the query should work in H2.
Anyone see what is wrong?
Thanks.

I don't believe FROM is allowed in the UPDATE syntax.

You can't update an alias, you need to have the table name specified.

Complementary to other answers, JOIN (just as FROM) is not allowed in UPDATE for H2. It would be allowed in a sub query.
Essentially, stick to the basic syntax:
UPDATE SomeTable as SomeAlias
SET SomeField = ?
WHERE (%GoWild%)
Whether or not you need the alias is up to your where clause.
Reference: http://www.h2database.com/html/grammar.html#update

Related

Problem in PostgreSQL (ERROR: relation "a" does not exist)

SQL query (PostgreSQL) looks like that:
UPDATE a
SET "PropertyAddress" = COALESCE(a."PropertyAddress", b."PropertyAddress")
FROM "NashvilleHousingData" a
INNER JOIN "NashvilleHousingData" b
ON a."ParcelID" = b."ParcelID"
AND a."UniqueID" <> b."UniqueID"
WHERE a."PropertyAddress" IS NULL;
And the error is relation "a" does not exist
I tried other advices about the notion in the code public or scheme, but it still doesn't work. Please, help
This is not how Postgres handles updates with more than one table. You don't repeat the table in the from clause:
UPDATE "NashvilleHousingData" nhd
SET "PropertyAddress" = COALESCE(nhd."PropertyAddress", nhd2."PropertyAddress")
FROM "NashvilleHousingData" nhd2
WHERE nhd2."ParcelID" = nhd."ParcelID" AND
nhd2."UniqueID" <> nhd."UniqueID"
WHERE nhd."PropertyAddress" IS NULL;
Also, the COALESCE() is superfluous because the value is known to be NULL:
SET "PropertyAddress" = nhd2."PropertyAddress"

SELECT instead of TABLE and WHERE in UPDATE?

I've seen may answers to the same kind of question but I still doubt.
UPDATE in SQL should be something like :
UPDATE *Table*
SET *choose value*
WHERE *what do we change*
I would like to know if there is possibilites to use a select instead of TABLE (an so instead of WHERE)
Like
UPDATE *Select conditions and rows*
SET *What do we change (the where is implicit)
I know UPDATE/SET/WHERE works well, but I'm exploring other possibilites :)
Thanks,
Nicolas
EXAMPLE :
Have to do :
update produits
set `NO_FOURNISSEUR` = "30"
where `NO_FOURNISSEUR` = "3"
would like to try something like :
update select * from produits where produits.`no_fournisseur`= "30"
set `NO_FOURNISSEUR`= "3"
MariaDB has a with expression, like so:
https://mariadb.com/kb/en/library/with/
so you would have:
WITH t AS (
select *
from produits
where produits.no_fournisseur= '30')
UPDATE t SET t.no_fournisseur = '3';
Yes. The ANSI SQL standard way to do this is using a Common Table Expression:
with dt as
(
select *
from produits
where produits.no_fournisseur= '30'
)
update dt set NO_FOURNISSEUR = '3'
This standard syntax supports joins and other query constructs in the SELECT part, and gives you a simple way to examine the rows before applying the update.

SQL Update on joined tables with calculated fields

First of all, I know there are already questions and answers about it, this thread being the one that is closest to what I need:
SQL Update to the SUM of its joined values
However, I get a syntax error (operator missing) that seems to occur close to the FROM clause. However I can't see it. Does it not like the FROM itself ? I am not used to using FROM in an update statement but it seems like it's valid from the QA I just linked :|
Any idea why there would be a syntax error there ?
I am using Access 2007 SP3.
Edit:
Wow, I forgot to post the query...
UPDATE r
SET
r.tempsmoy_requete_min = tmm.moy_mob_requete
FROM
rapports AS r INNER JOIN
(SELECT
id_fichier,
Round(Sum(temps_requete_min)/3,0) As moy_mob_requete,
Round(Sum(temps_analyse_min)/3,0) As moy_mob_analyse,
Round(Sum(temps_maj_min)/3,0) As moy_mob_maj,
Round(Sum(temps_rap_min)/3,0) As moy_mob_rap,
Round(Sum(temps_ddc_min)/3,0) As moy_mob_ddc
FROM maintenances
WHERE
periode In (10,9,8) And
annee=2011
GROUP BY id_fichier) AS tmm ON rapports.id_rapport = tmm.id_fichier
WHERE
1=0
The WHERE 1=0 part is because I want to test further the subquery before running it.
Edit: This is some simpler query I am trying. I get a different error this time. It now tells me that tempsmoy_requete_min (and probably all other left operands) are not part of an aggregate function... which is the point of my query. Any idea ?
UPDATE
rapports INNER JOIN maintenances ON rapports.id_rapport = maintenances.id_fichier
SET
rapports.tempsmoy_requete_min = Round(Sum(temps_requete_min)/3,0),
rapports.tempsmoy_analyse_min = Round(Sum(temps_analyse_min)/3,0),
rapports.tempsmoy_maj_min = Round(Sum(temps_maj_min)/3,0),
rapports.tempsmoy_rap_min = Round(Sum(temps_rap_min)/3,0),
rapports.tempsmoy_ddc_min = Round(Sum(temps_ddc_min)/3,0)
WHERE
maintenances.periode In (10,9,8) And
maintenances.annee=2011 AND
1=0
I tried adapting your first query sample, and was able to make your error go away. However then I encountered a different error ('Operation must use an updateable query').
It may be possible to overcome that error, too. However, I found it easier to use a domain function instead of a join to retrieve the replacement value.
UPDATE rapports
SET tempsmoy_requete_min = Round(DSum("temps_requete_min",
"maintenances",
"periode In (10,9,8) AND annee=2011 "
& "AND id_fichier='" & id_rapport
& "'")/3, 0);
If this suggestion works for tempsmoy_requete_min with your data, you will have to extend it to the other fields you want to replace. That won't be pretty. You could make it less ugly with a saved query which you then use as the "Domain" parameter for DSum() ... that could allow you to use a simpler "Criteria" parameter.
UPDATE r
should be
UPDATE rapports
You can't reliably use an alias in the update target.

SQL server syntax error in update statement, but I can't see it

I'm getting a syntax error on this query, but I can't figure it out.
Incorrect syntax near the keyword
'group'.
I believe its on the last group by, but I don't see whats wrong. Can anyone suggest how to correct this?
UPDATE [NCLGS].[dbo].[CP_CustomerShipTo]
SET TimesUsed = TimesUsed + B.NewCount
from [NCLGS].[dbo].[CP_CustomerShipTo] CST
INNER JOIN (
Select
PKH.CompanyCode,
PKH.CompanyName,
PKH.Addr1,
PKH.Addr2,
PKH.City,
PKH.State,
PKH.Zip,
Count(recid) As NewCount
from avanti_packingslipheader PKH
where pksdate > dbo.ufn_StartOfDay(DATEADD(d, -1, GETDATE() ) )
group by
PKH.CompanyCode,
PKH.CompanyName,
PKH.Addr1,
PKH.Addr2,
PKH.City,
PKH.State,
PKH.Zip
) B
ON CST.CustomerCode = B.CompanyCode
AND CST.ShipToName = B.CompanyName
AND CST.ShipToAddress1 = B.Addr1
AND CST.City = B.City
AND CST.PostalCode = B.Zip
group by
PKH.CompanyCode,
PKH.CompanyName,
PKH.Addr1,
PKH.Addr2,
PKH.City,
PKH.State,
PKH.Zip
BACKGROUND - I'm trying to do an update statement with a Count(), but of course you can't use agg. functions in an update set statement, so I'm trying to use a subquery.
You have already got GROUP BY inside the subselect, so what does the outer GROUP BY stand for?
You can't reference an alias in a subselect from an outer GROUP BY. But in any event you can't use GROUP BY with an UPDATE statement, and that's what the error message is about.
Try removing the last Group By. What exactly are you hoping this last group by will do?
Change the code to this:
update mytable set
mycolumn = mycolumn + (select x from ...);

UPDATE is not allowed because the statement updates view "table_name" which participates in a join and has an INSTEAD OF UPDATE trigger

I am getting the following error while executing the following query in an Stored Procedure. Could anyone help in finding the fault?
UPDATE is not allowed because the statement updates view "sup_item" which participates in a join and has an INSTEAD OF UPDATE trigger.
UPDATE si
SET
name = mc.name,
sup_item_cat_id = mc.res_sup_item_cat_id,
xf_value = mc.xf_value,
ava_start_date = mc.ava_start_date,
ava_end_date = mc.ava_end_date,
status_code = mc.status_code,
last_mod_us_id = CASE WHEN mc.last_mod_us_id = 42 THEN #posting_us_id
ELSE mc.last_mod_us_id END,
last_mod_tsp = CURRENT_tsp
FROM sup_item AS si
JOIN merch_cat_imp_sup_item AS mc
ON mc.sup_id = si.sup_id
AND mc.res_sup_item_id = si.sup_item_id
AND mc.cat_imp_event_id = #cat_imp_event_id
AND mc.accept_flag = 'y'
WHERE si.shi_flag = 'n'
I found the reference: http://msdn.microsoft.com/en-us/library/ms177523.aspx
A view with an INSTEAD OF UPDATE trigger cannot be a target of an
UPDATE with a FROM clause.
So, I have to rewrite the UPDATE statement (it still can be in a procedure) to NOT use sup_item (which is a view), but keep the underlying table(s) as needed.
Could someone please rewrite it, if anyone knows what to do?
You can use MERGE to achieve this. Try:
MERGE INTO sup_item si
USING merch_cat_imp_sup_item AS mc
ON mc.sup_id = si.sup_id
AND mc.res_sup_item_id = si.sup_item_id
AND mc.cat_imp_event_id = #cat_imp_event_id
AND mc.accept_flag = 'y'
AND si.shi_flag = 'n'
WHEN MATCHED
THEN UPDATE
SET
name = mc.name,
sup_item_cat_id = mc.res_sup_item_cat_id,
xf_value = mc.xf_value,
ava_start_date = mc.ava_start_date,
ava_end_date = mc.ava_end_date,
status_code = mc.status_code,
last_mod_us_id = CASE WHEN mc.last_mod_us_id = 42 THEN #posting_us_id
ELSE mc.last_mod_us_id END,
last_mod_tsp = CURRENT_tsp
The issue is not within your query. As per comments on your question, the entity you are updating [sup_item], isn't actually a table, it's a view. That view has an INSTEAD OF UPDATE trigger on it.
Are you able to post the SQL for the View and for the Trigger(s)?
I would also be interested, because I have a stored procedure in a database that I have inherited which tries to do this. It won't let me create the sproc in SQL 2014, but the fact that it is there in the sproc indicates to me that an earlier version of SQL server must have allowed this.
Maybe in earlier versions your procedure operated on a table, which was later replaced by a view.
You should replace your "update from" syntax by standard ANSI syntax of update.