# Controllers

Istio has a variety of [controllers](https://kubernetes.io/docs/concepts/architecture/controller/), which basically watch some inputs and do something.
This can be reading from Kubernetes and writing other objects back, writing to proxies over XDS, etc.

Unfortunately, writing controllers is very error prone, even for seemingly simple cases.
To work around this, Istio has a variety of abstractions meant to make writing controllers easier.

## Clients

		requestorIsDerivedCredential = true
	}

	// Check if we are creating svc account for request sender.
	isSvcAccForRequestor := false
	if targetUser == requestorUser || targetUser == requestorParentUser {
		isSvcAccForRequestor = true
	}

	var (
		targetGroups []string
		err          error
	)

	// If we are creating svc account for request sender, ensure that targetUser

	db.Stmts[query] = &cacheStmt
	db.Mux.Unlock()

	// prepare completed
	defer close(cacheStmt.prepared)

	// Reason why cannot lock conn.PrepareContext
	// suppose the maxopen is 1, g1 is creating record and g2 is querying record.
	// 1. g1 begin tx, g1 is requeue because of waiting for the system call, now `db.ConnPool` db.numOpen == 1.

        int n;
        if(( n = readn( in, buffer, bufferIndex, HEADER_LENGTH )) != HEADER_LENGTH ) {
            if( n == -1 ) {
                return -1;
            }
            throw new IOException( "unexpected EOF reading netbios session header" );
        }
        int t = buffer[bufferIndex] & 0xFF;
        return t;
    }

    int type, length;

    public int writeWireFormat( byte[] dst, int dstIndex ) {

				totalGet += result.Downloads
			}

			if totalGet < throughputHighestGet {
				// Following check is for situations
				// when Writes() scale higher than Reads()
				// - practically speaking this never happens
				// and should never happen - however it has
				// been seen recently due to hardware issues
				// causes Reads() to go slower than Writes().
				//
				// Send such results anyways as this shall

                long lm;

                if ( ( lm = f.lastModified() ) > this.lastModified ) {
                    if ( log.isDebugEnabled() ) {
                        log.debug("Reading " + tc.getConfig().getLmHostsFileName());
                    }
                    this.lastModified = lm;
                    this.table.clear();
                    try ( FileReader r = new FileReader(f) ) {

              return new String(buffer, 0, bufIndex);
            }
            // otherwise we got the size wrong.  This can happen if the size changes between when
            // we called sizeIfKnown and when we started reading the file (or I guess if
            // maxCharsPerByte is wrong)
            // Fallback to an incremental approach
            StringBuilder builder = new StringBuilder(bufIndex + 32);

Calls can be canceled from any thread. This will fail the call if it hasn’t yet completed! Code that is writing the request body or reading the response body will suffer an `IOException` when its call is canceled.

## Dispatch

    # this if using obj.model_validate becomes independent of Pydantic v2
    with warnings.catch_warnings(record=True):
        warnings.simplefilter("always")
        # No heroes before creating
        response = client.get("heroes/")
        assert response.status_code == 200, response.text
        assert response.json() == []

        # Create a hero
        response = client.post(
            "/heroes/",

    prefixes instead of exact match.
```

- There is always a possibility of a fix that is new, it is advised that the developer must make sure that the fix is sent upstream, reviewed and merged to the master branch.

## Creating a hotfix branch