I have a property, Emails, that needs to contain a list of comma-delimited values, or "". Is there a benefit to making this property a list of %String, a Collection[list] (or however you do it), an array, or anything other than a simple %String type?

I just  need to store some values, and return these values in SQL -- so it needs to be SQL compatible.  I don't need to index the values or use individual values in the SQL statement (e.g. Select * from table where emails [ "email@co.com"  -- I don't need to do this, although maybe this would work anyway if it's of type %String).

We've been noticing that we're getting this (https://community.intersystems.com/post/cach%C3%A9-sql-error-odbc) error more frequently and having to constantly purge cached queries in order to resolve it. Is there a way to avoid the issue in the first place?

thanks!

I have the following query which tells me how many documents were retrieved for each customer, but it only works for the "on-demand" customers:

All the other PatientFacility's show up as "Documents".

I want to override getter for a serial property as a whole, because getter code depends on a class in which a property is defined.

Let's say I have CS.Serial as a serial class:

Class CS.Serial Extends %SerialObject
{
Property Year As %Integer;

Property Month As %Integer;
}

And CS.Persistent as a persistent class:

Class CS.Persistent Extends %Persistent
{

Property data As CS.Serial [ SqlComputeCode = {set {*} = ##class(CS.Persistent).dataGetStatic()}, SqlComputed, Transient ];

/// data getter method in object context
Method dataGet() As CS.Serial
{
    return ..data2GetStatic()
}

/// data getter method in SQL context
ClassMethod dataGetStatic() As CS.Serial
{
    return $lb($lb("2017","01"),"CS.Serial")
}
}

So I know it's been a while, and I hate to let my adoring fans down... just not enough to actually start writing again.  But the wait is over and I'm back!  Now bask in my beautiful ginger words!

For this series, I am going to look at some common problems we see in the WRC and discuss some common solutions.  Of course, even if you find a solution here, you are always welcome to call in and expression you gratitude, or just hear my voice!

This week's common problem: "My query returns no data."

When working with a large query executed though an ODBC connection what is the best way to allow the paging of the results at the client side.  I have tried some methods using %VID and similar methods, but these really don't seem to work as the value returned is related to the ID of the data and not the position in the results set.  What would be ideal is if the value seen in the management portal when you check of "Row Number" was available to external queries through ODBC.  I have not seen a way to return this however.   This would be similar to SQL server Last n and Skip n capabilities.

Hi guys.

Do You known If Caché has suport to Common Table Expressions (With queries)

By example: https://www.postgresql.org/docs/9.6/static/queries-with.html

Hello,

I read the Cache Documentation where it describes Private Property as below:

If you have a lot of SQL queries to external systems (or even internal dynamic ones), you can encounter something like this:

Set SQL =    "SELECT "_
                "c.cid                        AS Id,"_
                "c.nid                        AS Nid, "_
                "FROM_UNIXTIME(c.created)     AS Created,"_
                "c.uid                        AS Uid,"_
                "IFNULL(vv.average,0)         AS AvgVote,"_
                "IFNULL(vv.amount,0)          AS VotesAmount,"_
                "body.comment_body_value      AS Text,"_
                "'comment'                    AS Type "_
            "FROM comment AS c "_
                "LEFT OUTER JOIN node ON node.nid = c.nid "_  
                "INNER JOIN field_data_field_forum_ref AS ref ON ref.entity_id = node.nid "_
                "LEFT OUTER JOIN field_data_comment_body AS body ON c.cid = body.entity_id "_
                "LEFT OUTER JOIN (SELECT entity_id , SUM(value) AS average, COUNT(1) AS amount "_
                                  "FROM votingapi_vote "_
                                  "WHERE entity_type = 'comment' "_
                                  "GROUP BY entity_id) AS vv ON vv.entity_id = c.cid "_
                ....
                + 100 more lines of SQL code omitted, but you get the idea

So, I played with row level security and was unable to find a way to get to work it dynamically. I want to determine if the user can access the row in runtime based on a state of external system, but it seems that row level access is calculated during INSERT/UPDATE and stored in %RLI index. Is there a way to achieve runtime access calculation?

Hello,

Is there some way to reproduce the SQL GROUP_CONCAT(http://sql.sh/fonctions/group_concat) with the Caché DB ?

 That is exactly what I would expect.

That is NOT expected. 

We don't often use SQL within our org, which is mostly due to the performance issue we experience due to the quantity of data we are reviewing.

Aside from the standard performance measures for non-Caché databases, are there any recommended approaches when querying large tables?

The table would have roughly 50M records, but there are not a finite amount of sub-nodes.

Hello Fellow Caché Developers,

The purpose of this post is to ask for everyone's thoughts and input around the use of transient, SQLCalculated/Computed properties within persistent classes.

This approach allows for extra data values needed in SQL queries to be available without having to join to other tables.

Very simple/basic example to illustrate the topic:

Let's say I have a persistent class named ICDAutoCodeDefn to hold ICD Auto-Coding definitions, such as:

Class ApplicationName.DB.MedicalCoding.

 Error:

    [SQLCODE: <-400>:<Fatal error occurred>]
    [Cache Error: <<SUBSCRIPT>%0AmEdun+4^%sqlcq.HSREGISTRY.cls966.1 ^||%sql.temp(1,"")>]
    [Location: <ServerLoop - Query Fetch>]
    [%msg: <Unexpected error occurred:  <SUBSCRIPT>%0AmEdun+4^%sqlcq.HSREGISTRY.cls966.1 ^||%sql.temp(1,"")>]

Here is an example of a query that gives the error:

    SELECT COUNT(DISTINCT Criteria) as Relevance FROM HS_IHE_ATNA_Repository.Aggregation
    WHERE EventType = 'CROSS GATEWAY QUERY'

The content of the actual field is XML, but the query simply returns a number.

I have a class which defines a property as array of %String. Is it possible to index values of this property and use this property in SQL?

I have tried 'Index idx On prop(ELEMENTS)' and then a select from the generated collection table, but this is still orders of magnitude slower than queries to the containing class.

Is there a tool to get the slowests queries ?

In writing some code on:

Cache for Windows (x86-64) 2015.1 (Build 429U) Fri Feb 13 2015 14:37:23 EST

I noticed an unexpected "ROLLFAIL" error when a Unique Index fail was generated in a %Save() call.

The object SHOULD fail to save, since the Unique criteria is not met, but I didn't expect to see a ROLLFAIL error as the "Last Error".

When running the following class code:

Class BUG.Test Extends %Persistent
{

Property Field1 As %String;

Property Field2 As %String;

Index idx1 On Field1 [ Unique ];

ClassMethod Test()
{
    kill ^BUG.TestD
    kill ^BUG.TestI

    ; Should work fine
    set record1 = ##class(BUG.Test).%New()
    set record1.Field1 = "Bosco"
    set record1.Field2 = "Jones"
    set sc = record1.%Save()
    if $$$ISERR(sc) {
        do $System.OBJ.DisplayError(sc)
    } else {
        write !,"Save OK ID: "_record1.%Id()
    }

    ; Should work fine
    set record2 = ##class(BUG.Test).%New()
    set record2.Field1 = "Waffles"
    set record2.Field2 = "Syrup"
    set sc = record2.%Save()
    if $$$ISERR(sc) {
        do $System.OBJ.DisplayError(sc)
    } else {
        write !,"Save OK ID: "_record2.%Id()
    }

    ; Should fail on duplicate "unique" value for Field1
    set record3 = ##class(BUG.Test).%New()
    set record3.Field1 = "Bosco"
    set record3.Field2 = "Jones"
    set sc = record3.%Save()
    if $$$ISERR(sc) {
        do $System.OBJ.DisplayError(sc)
    } else {
        write !,"Save OK ID: "_record3.%Id()
    }
}

I would like to integrate SQL into our ObjectScript routines in order to illustrate how SQL can be used to minimize development time. However, there are two questions I have.

1. How does one iterate through the results of a SELECT statement?
2. How does one access a Table in another namespace?

I'm trying to use the standard TSQL DECLARE syntax but I get the following error:

[%msg: < IDENTIFIER expected, @ found^  DECLARE @>]

How do I declare a variable?  For instance:

i have this error

DrawTableError : <MAXSTRING> zDrawTable+349^%CSP.Util.TablePane.2

perhaps i have the possibility of Deleting all history queries, but i think better to check the table where last queries were stored and solve it by deleting wrong registries. anybody knows where is the table 'QueryHistory'.

I get this on some queries in some namespaces.  For instance, this query:

It works fine in HSBUS but in HSREG it throws the error.

Server closed communication device

Does anyone know what would cause this? Would it log something more useful somewhere?

Since now i have been working from external connection but i want to work with the SQL utility of Management Portal

i dont know how to do with several instructions like in other editors like this example

update Prod.Articulos set Alto = 1646, Ancho = 16, Fondo =  80 where CodigoNum = '100' and Empresa = 'CO'
update Prod.Articulos set Alto = 1646, Ancho = 16, Fondo =  400 where CodigoNum = '101' and Empresa = 'CO'
update Prod.Articulos set Alto = 1646, Ancho = 16, Fondo =  400 where CodigoNum = '102' and Empresa = 'CO'

Running TuneTable accounts among other the parameter named Block Count. In documentation, we see that this is an approximate numbers of 2K-blocks in which SQL-maps are stored. Databases in recent Cache doesn't support 2K-physical blocks so SQL-blocks are not physical blocks as it seems. So two questions:

- what are these blocks?

- how knowledge about count of blocks can help in SQL optimization?

Thanks for intelligent answer to stupid questions!

Beginning in Caché 2013.1, InterSystems introduced Outlier Selectivity to improve query plan selection involving fields with one atypical value.

In this article, I hope to use an example 'Projects' table to demonstrate what Outlier Selectivity is, how it helps SQL performance and a few considerations for writing queries.

Selectivity

First, let's take a quick look at Selectivity. Selectivity is meta information about the values in one column in a table.

I've asked a lot of questions leading up to this, so I wanted to share some of my progress.

The blue line represents the number of messages processed.  The background color represents the average response time.  You can see ticks for each hour (and bigger ticks for each day).   Hovering over any point in the graph will show you the numbers for that period in time.

This is super useful for "at a glance" performance monitoring as well as establishing patterns in our utilization.

Here is the query used:

SELECT
    mh.name                                  AS MessageType,
    COUNT(mh.name)                           AS MessageCount,
    CAST(AVG(ResponseTime) AS DECIMAL(5, 2)) AS AvgResponseTime
FROM
    (
     SELECT
            li.SessionId,
            li.Name,
            DATEDIFF(s, MIN(li.TimeCreated), MAX(lo.TimeCreated)) AS ResponseTime
       FROM
            (
             SELECT
                    SessionId,
                    name,
                    TimeCreated
               FROM
                    ens.messageheader h1,
                    HS_Message.XMLMessage m1
              WHERE
                    h1.MessageBodyId = m1.ID
                AND h1.TimeCreated > DATEADD(hh, -1, GETUTCDATE())) li
       JOIN
            (
             SELECT
                    SessionId,
                    TimeCreated
               FROM
                    ens.messageheader h2,
                    HS_Message.XMLMessage m2
              WHERE
                    h2.MessageBodyId = m2.ID
                AND h2.TimeCreated > DATEADD(hh, -1, GETUTCDATE())) lo
         ON
            li.SessionId = lo.SessionId
   GROUP BY
            li.SessionId) mh
WHERE
    mh.name LIKE '%REQUEST'
GROUP BY
    mh.name

This post is the direct result of working with an InterSystems customer who came to me with the following problem:

SELECT COUNT(*) FROM MyCustomTable

Takes 0.005 seconds, total 2300 rows.  However:

SELECT * FROM MyCustomTable

Took minutes.  The reason for this is subtle and interesting enough for me to write a post about.  This post is lengthy, but if you scroll to the bottom I'll write a quick summary, so if you've gotten this far and think you've already read enough, scroll to the end to get the main point.  Check for the sentence in bold.

Hi All,

            I have a general query in regards to developers experience on extracting data from cache databases and the most efficient way to do so. I work with a number of clients who have applications with cache databases and require the data off the host system and onto data warehouse platforms for research and analysis. Often they require the data in source state which means the extracts are often simply a table scan of the entire database table without any aggregation or manipulation.

This tells me that there is no timezone offset on this table/field:

So why is TimeCreated in this resultset 4 hours less than OneMinuteAgo when I'm clearly requesting only records with a TimeCreated greater than OneMinuteAgo?

I have a list of about 100 MPI IDs that I would like to run a report on.  I want to list times that any data for these patients were accessed.  Currently in "Managed Reports" we have a "Disclosure Report" which I think was a custom development effort, but it is per-patient.

I have a SQL query for the ATNA log but I'm not confident in its accuracy, so I thought I'd reach out and see how other Information Exchange's might get this data.