The README gives very simple examples like
Response response = db.remove(foo);

However it does not say what is supposed to happen in case of issues, such as a document not found, or any other error for any method.

Can someone clarify that?
Exceptions? Null response? Response with some error property?

How to check whether the request was completed successfully or not...

Hi, I'm trying to figure out if it is possible to run a query for a particular key-value-pair with a customized sort setting and then in the returned doc it will returns the index of the doc in the database along with the doc. Do you think it is doable? Thanks.

For example:
I have a database that has 10 docs, and I am looking up the one with name=john, with sorting on address. I wish Cloudant can return me the doc that has name=john while also letting me know its index in this data set with sorting applied is 5.

make CouchDbClient class have only Account/Server related methods & all DB related methods move to a new class 'CouchDatabase'.

Could a check for wsjar be added, in addition to jar and file, to the org.lightcouch.internal.CouchDbUtil.listResources(String path) method. I am using WebSphere Liberty 8.5.5.5 and that method returns null for a path like the following:
wsjar:file:/somepath/myapp.war!/WEB-INF/classes/design-docs/my-filter/

@mikerhodes
1- any ideas on how to test this one. Will we get a dedicated/sharded account?
2 - Should i just make available 2 properties on the DB object, rather than adding a 'readQuorum/writeQuorum' paramter to every read/update method? Do we have any precedents here ?

I tried calling db.view("_all_docs").includeDocs(true).query(JSONObject.class) on my code but I'm getting weird errors about Gson, and it raised an exception:
java.lang.IllegalArgumentException: Invalid type of value. Type: [com.google.gson.internal.LinkedTreeMap] with value: [my_json_but_with_all_quotes_stripped]

When I called the function with Object.class instead of JSONObject.class, I got an Object which, when printed, looks almost like JSON but with all of the quotes stripped out (even for strings with whitespace).

When I called the function with String.class, I got some error along the lines of "Expected STRING but found BEGIN_OBJECT".

Is there a way to get the new _rev field using the invokeUpdateHandler method to invoke an in-line update in Cloudant?
Actually the class com.cloudant.client.api.Database expose only the method public String invokeUpdateHandler that returns the response stream of the update operation. But the new _rev field is in the header of the http response (https://wiki.apache.org/couchdb/Document_Update_Handlers#Response).
So, is there a way to get it?

The README.md, CHANGES.md and javadoc need to be slimmed down and up-to-date for a 2.0 release.

The API reference should be moved entirely into javadoc so the examples don't need to be maintained in two places.
A javadoc overview.html and some package-info.java classes are needed.
There are also a lot of javadoc warnings that need resolution so we can maintain the javadoc in a clean state.

We started seeing memory run ups after putting the library under low load over a number of hours. The datasets are not trivial, maybe 50k docs with 20 fields.

But the question is really about using the cloudant client efficiently.
Is it best to use a single singleton instances across a tomcat app?
Or should we create a pool of them?
Or just create as needed?
Always call shutdown (i saw apache http client is used underneath which recommends cleaning up.)
Thanks.

Suggest just pointing to the top level repo so that you don't have to keep updating the location of the jar file. That is, https://repo1.maven.org/maven2/com/cloudant/cloudant-client/ vs. https://repo1.maven.org/maven2/com/cloudant/cloudant-client/1.0.1/cloudant-client-1.0.1.jar

After replacing Apache HttpClient with HttpUrlConnection connection pooling is relying on the JVM wide persistent connection settings.
It would be preferable to use the previous style of configuring connection pooling on a per CloudantClient instance basis.

I am currently getting a java.lang.IllegalArgumentException: AuthCookie may not be null.

Here is my code

public class Cloudant {
public static void main (String[] args){

String password = System.getProperty("gffgasdas");
CloudantClient client = new CloudantClient("test1234.cloudant.com",password);

System.out.println("Connected to Cloudant");
System.out.println("Server Version: " + client.serverVersion());

List<String> databases = client.getAllDbs();
System.out.println("All my databases : ");
for ( String db : databases ) {
    System.out.println(db);

Hello,
I have a view that returns complex keys that follow this format/pattern:

["1000","2000”,0]
["1000","2000",1]
["1000","2000”,2]
["1000","2001”,0]
["1000","2001”,1]
["1000","2001”,2]
["1001”,”2000”,0]
["1001”,”2000",1]
["1001”,”2000”,2]
["1001”,”2001”,0]
["1001”,”2001”,1]
["1001”,”2001”,2]
...

I can query directly my view [using curl from the command line] by submitting a REST request and specifying the following start and end key values:

startkey=["1000","0",0]&endkey=["1000",{}]"

The above start and end key values return the data I am expecting. Basically, I am filtering the results and getting only those records that have "1000" as the first value in the complex key.

However, when I try to run the same query against the view using the Java API, I don’t get the expected results:
...
Object[] startKeys = new Object[] { "1000", "0", 0 };
Object[] endKeys = new Object[] { "1000", “{}” };
view.startKey(startKeys);
view.endKey(endKeys);
....

I get no results back when using the start and end key values shown above in the Java code. I am wondering if there is a defect in the API code that builds the URL to query and filter the view. Could you provide some guidance on how to resolve this issue? Thanks.

http://docs.cloudant.com/api/cloudant-query.html - The entire thing

DesignDocoument.getFromDesk expects views to be in a specific sub folder of the current working directory. The folder is called design-docs. This is not spelt out in the documentation and seems to be an arbitrary restriction on where the library will load views from. It should be replaced with DesignDocument.fromFile(...) or something similar.

i.e. https://username.cloudant.com/_api/v2/api_keys & https://username.cloudant.com/_api/v2/db//_security support AuthSession

http://docs.cloudant.com/api/misc.html
_membership

The findByIndex() API takes a 'selector' which the query docs give as a JSON object such as

{"name": "Paul"}

however the README gives the example:

List<Movie> movies = db.findByIndex("\"selector\": 
                { \"Movie_year\": {\"$gt\": 1960}, \"Person_name\": \"Alec Guinness\" }",

.. where the selector is more of a "fragment" than a well formed object. In fact, I found that the server returned JSON errors unless I called findByIndex with "\'selector\':" + someJson (where someJson is the well formed JSON selector body).

Could this API be clarified, or even better accept just the JSON body and not require the key? Otherwise it is awkward to create this "Fragment" for use with the API.

The dbcopy test can fail intermittently due to eventual consistency when running against the Cloudant service. It is nice to assert that the correct number of documents were copied, so we can solve this by adding a retry loop.

com.cloudant.client.api.View.queryView(Class k, Class v, Class t) can use some more docs. In particular, t is the type of the doc, k the type of the key, and v of the value. This is handy when querying something that has a complex key.

The java-cloudant client should use the same URI builder as sync-android. A number of fixes have gone into that URI builder for special cases (e.g. ":") and we should take advantage of them. Additionally it is another step towards making java-cloudant a suitable base for sync-android.

Creating a CloudantClient instance from a URL without specifying a port results in the port being set to -1 which is invalid e.g.

new CloudantClient("http://192.0.2.0", "user", "pass");

It is possible to workaround by specifying the port e.g.

new CloudantClient("http://192.0.2.0:80", "user", "pass");

It shouldn't be necessary to specify the default port.

http://docs.cloudant.com/api/search.html - The entire thing

Hi,

I found a problem with changes stream, even in the rest api. It's not returning the same results than the cloudant dashboard. I have tried the style("all_docs") option but it's returning the same.

Also, I have found that sometimes the result is different while calling the changes endpoint through the rest api : https://xxx.cloudant.com/dashboard.html#/database/yyy/_changes?

Thanks

Building JSON literals using string operations is dangerous. Database.getIndexDefinition() is the first place I spotted this but there may be other examples in this codebase. GSON is a better choice.

@mikerhodes
1 - looking at /db/_shards here (http://docs.cloudant.com/api/database.html), I am assuming the # of shards will be variable i.e 00000000 to ffffffff can be split over as many shards as required. So wouldnt 'shards' be an array?... having problems mapping to a domain object if the set of ranges are not fixed and then it is not an array.

2- What about the endpoints after this one in http://docs.cloudant.com/api/database.html i.e. starting with 'Cleaning up cached view output'
If yes, is there some overlap between 'set_permissions' and 'PUT /db/_security'

There is a bug in < http client 4.3.6 that allows a read to block forever even when socket timeout is set.

Hi

I'm trying to set breakpoints in the Cloudant lib code from within IntelliJ IDEA, but it's not working. This can happen if a jar is built but no debug information is included. If this is a case, could you release a new jar version that includes debug information?

Thanks!

The Response class does not include the HTTP response code. We should add this information to help people align client behaviour to the more general Cloudant documentation, especially useful for cases like 201 and 202.

We should validate that the status code has been added to CouchDbException and its subclasses where possible as well.

An in-memory or other cache could offer performance benefits for some client use cases.

Hi,

Please update maven dependency jar(1.0.1) with new code changes.
I am not getting keys(Object... keys) changes with dependency jar.

Old Issue link:
#42

Thanks,

The Cloundant query sample snippet code below is not correct:

List<Movie> movies = db.findByIndex("{
                     \"Movie_year\": {\"$gt\": 1960}, \"Person_name\": \"Alec Guinness\"
                     }",
                    Movie.class,
                    new FindByIndexOptions()
                        .sort(new IndexField("Movie_year", SortOrder.desc))
                        .fields("Movie_name").fields("Movie_year"));

I tried something similar:

String Selector="{\"Place_id\": {\"$gt\": 100}}";
   List<places> p1 = db_use.findByIndex(Selector,places.class);

and got a JSON exception:

org.lightcouch.CouchDbException: Bad Request{"error":"bad_request","reason":"invalid UTF-8 JSON"} 

After adding "selector" to the query it then worked:

String Selector="\"selector\": {\"Place_id\": {\"$gt\": 100}}";

Per cloudant/nodejs-cloudant#6 (comment) this will handle the merging into /dbname/_security

see http://docs.cloudant.com/api/authz.html

It seems that a view can only be queried with a single key, using the key(Object...) method of com.cloudant.client.api.View. In the meantime, cloudant allows querying using multiple keys (using query parameter keys=...), so it would be helpful to have a method that allow querying using multiple keys (for example a method keys(Object[]) on the com.cloudant.client.api.View class).

I think there's a bug with the startKey and endKey functions as well as key()

Even in the ViewsTest there is this snippet in byComplexKey()

    public void byComplexKey() {
        int[] complexKey = new int[] { 2011, 10, 15 };
        List<Foo> foos = db.view("example/by_date")
                .key(complexKey)
//              .key(2011, 10, 15) 

I think the problem is that one (Object...) function is calling another (Object...) function. I can't remember how to fix this yet, but want to capture.

I had to do similar code to the above workaround in my own project.

CloudantClient(String account, String loginUsername, String password)

• The second parameter here could be documented as "apikey or account" if I understand properly: "If using an API key, this is the api key, otherwise pass in the account for this parameter also" ?
• Actually, could you take a "null" loginUsername here and then: if loginUsername==null { loginUsername = account} ? Maybe that would be more confusing than helpful, though.

Please include documentation how to setup and configure connection pooling for this library. For example:

1. information on instantiating the library so that it can be shared, e.g. using a singleton
2. calling out which methods are threadsafe and which aren't
3. describe how to configure connection pooling
4. describe how the library authenticates to Cloudant (using cookie or basic authentication)
5. describe if session expiry gets handled automatically by the library [1]. if not, what does the library consumer need to do to handle this?
6. how to use the library in JEE environments so that connection pooling can be configured at the app server level [2]

There is some information in this ticket: #34. However, it would be good if the README.md covered the above info.

[1] this assumes the library uses cookie authentication. The docs state that the cookie can be used until it expires.
[2] presumably this will require wrapping the library in a technology like JCA - [more info]

The POST to _session for the cookie auth in 2.0 needs to URL encode the username and password fields.

This is maybe a lightCouch q, but I wonder if synchronizeWithDb could be made more efficient with something such as:

• md5 hash the document, do a query including that hash?
• keep a 'revision number' in the js file (manually updated) and only sync if it changes?

I therefore get the following when trying to run on Java 6:

Exception in thread "main" java.lang.UnsupportedClassVersionError: com/cloudant/client/api/CloudantClient : Unsupported major.minor version 51.0

Is there any way in this API to POST to an update handler? Passing in no docId just results in an error docId is required where I would expect it to POST instead of PUT.

java.lang.IllegalArgumentException: docId may not be null.
    at org.lightcouch.internal.CouchDbUtil.assertNotEmpty(Unknown Source)
    at org.lightcouch.CouchDatabaseBase.invokeUpdateHandler(Unknown Source)
    at com.cloudant.client.api.Database.invokeUpdateHandler(Unknown Source)

Every request and response from the database is logged at INFO level. Can this be changed to DEBUG please.

log.debug("> " + req.getMethod() + " " + URLDecoder.decode(req.getUri(), "UTF-8"));
log.debug("< Status: " + response.getStatusLine().getStatusCode());

We should consider removing these methods and calling the ones in CouchDbClient:
<T> T get(URI uri, Class<T> classType)
InputStream get(URI uri)
Response put(...)

Hello, I am having an issue when using the query method of the View class:

List images = myView.query(Image.class);

The above method invocation results in a JSON exception: my Image class has a timestamp instance variable, and the Gson instance used internally by the couchdb library fails to translate a string like “2015-01-23T18:25:43.511Z” to a timestamp variable.

If I could somehow configure the Gson instance used internally by the couchdb library, this exception would not occur. I understand I can configure my own Gson instance in the following way to handle timestamp strings like “2015-01-23T18:25:43.511Z” correctly:

Gson gson = new GsonBuilder().setDateFormat("yyyy-MM-dd'T'HH:mm:ss").create();

However, I am afraid I don’t know how to configure the Gson instance used internally by the couchdb library. Could you provide some guidance on how to address this issue? Thanks.