27 Oct
2017
27 Oct
'17
2:43 p.m.
I'm trying to come up with an application that provides:
1)Ability for a user on a desktop to communicate with a mainframe, without connecting the browser directly to the mainframe code. I do not want to run a web server on the mainframe due to the security issues I will hit with that. This has to be an 'outbound' connection from the mainframe. The mainframe establishes the connection, does not listen for and accept a connection.
2)Ability for the mainframe to create and load databases within Basex. I tried establishing a direct connection to a Basex server, but that requires my mainframe code to have a crypto service running that will create the MD5 hash value. If that service isn't up, I can't get past the connection, and I can't require that the service be started.
3)Ability for a user to view the data in the database(s) as they are being updated to get the most current data.
My thought was to have the java comm server start, establish the RMI environment, and then when a command from the user came thru that interface, the java comm server would also start a new thread to connect to the Basex server and that thread would then create and load the database.
It would be a more simple and straight forward application if I could get the user's browser to communicate directly with the java comm server, and I am trying to research how to do that now.
My biggest issue with doing all of this is that I'm not in my own element. I'm having to teach myself java on the fly, as I go.
-- Dave
On 10/27/2017 1:01 PM, Kendall Shaw wrote:
Hi,
Do you need to be doing remote procedure calls? In the abstract possibly anything you would want RPC for could be done in java on the communications server and basex would be interacted with using it’s APIs.
I have been giving what I think are answers to abstract questions. It might be a good idea to ask more high-level questions that allow people to offer possible solutions using XQuery XML, XML databases, web applications, etc. Describing what you want a solution for that you think would require an XML database, might be useful.
Kendall
On 10/27/17, 8:57 AM, "basex-talk-bounces@mailman.uni-konstanz.de on behalf of Dave Day" <basex-talk-bounces@mailman.uni-konstanz.de on behalf of David.Day@duke-software.com> wrote:
Kendall, No, not wanting to use RMI to call functions in Basex, but to handle communications between a user on a browser, and my java comm server that is talking to my code on the z/OS box. Something like this: Java comm server starts up, and listens for a connection. Connection established, and a new thread is started. New thread checks to see if Basex is started, and starts one if not. Thread establishes an RMI server. When Basex starts, it starts my RMI client code. Would have to run on a separate thread within Basex. (Haven't figured out how to do this one yet) User, on a browser, while connected to the Basex server, needs to communicate with the mainframe code. Basex passes the request to the RMI client, then waits for the response from the RMI client.(Ditto on this one too...need to figure out if this is possible). RMI client sends request to RMI server thread in my comm server. RMI server thread pushes request across connection to mainframe. Waits for response. Response come in, RMI server thread pushes response to RMI client running in Basex. RMI client pushes response to Basex that then responds to the browser. User views the response. On 10/27/2017 10:34 AM, Kendall Shaw wrote: > HI, > > I would think so. But, if you mean using RMI to call functions in BaseX, it’s probably a better idea to run basex on both and use the BaseX client API or REST. > > Kendall > > On 10/27/17, 8:13 AM, "basex-talk-bounces@mailman.uni-konstanz.de on behalf of Dave Day" <basex-talk-bounces@mailman.uni-konstanz.de on behalf of David.Day@duke-software.com> wrote: > > Kendall, > > I've now got a java server running that can accept a connection > from my mainframe code. > > In the scenario you have below, would it be possible to have the > imported module use RMI to communicate with the java communication server? > > Regards, > > -- Dave > > > > On 10/19/2017 8:48 PM, Kendall Shaw wrote: > > Yes. The section in the documentation about RESTXQ shows how XQuery can be used to handle HTTP requests. The part about REST shows a way to execute queries (programs) and commands over HTTP. BaseX comes with a servlet that can run in a web server. The servlet is a java class that is executed by the java vm that is running the web server. The servlet also runs the native basex server. BaseX also includes a web server that can be used instead of a web server outside of BaseX. > > > > The section in the documentation about Java bindings shows how you can create modules that implement XQuery functions in java (or whatever JVM language). In xquery like: > > > > Import module namespace d = “ http: //duke-software.com/xml “; > > > > d:compute() > > > > The function d:compute() could be implemented in java so you are executing java by invoking the function. > > > > The XQuery that implements a web application could use the module that includes the java functions that interact with the mainframe to communicate with a service listening in the mainframe. That is driven by the web application. > > > > To cause basex to handle communication from the second task, you could use threads but instead of what I said in the last post, it might be better to have the program that interacts with the mainframe run as a separate OS service that handles data coming from the mainframe and data coming from basex. > > > > The user’s browser calls the web service that runs within basex (directly or from another web application). That web service includes calls to xquery functions that directly communicate with the mainframe over sockets or the separate OS service. > > > > The second task could communicate with the OS service that executes XQuery in basex and receives data from BaseX which it send to the mainframe. > > > > If you implement xquery functions in java you probably would need to look at the examples in the basex source in the basex-examples directory. > > > > Kendall > > > > On 10/19/17, 2:24 PM, "basex-talk-bounces@mailman.uni-konstanz.de on behalf of Dave Day" <basex-talk-bounces@mailman.uni-konstanz.de on behalf of David.Day@duke-software.com> wrote: > > > > It is your last sentence that I don't understand, due to my lack of > > knowledge of most all things Basex. > > > > The Java code that does the communication would be a function that is > > called from within a running Basex server? > > > > -- Dave > > > > > > On 10/19/2017 4:12 PM, Kendall Shaw wrote: > > > If the desktop code can be java, then it could be like: > > > > > > A web application running in basex serves the web page that is UI for the user. > > > > > > That web application uses the module that you wrote in java to connect and send and receive data with the mainframe for your 2 tasks. The module is used in the web application as xquery functions. > > > > > > Kendall > > > > > > On 10/19/17, 1:57 PM, "basex-talk-bounces@mailman.uni-konstanz.de on behalf of Dave Day" <basex-talk-bounces@mailman.uni-konstanz.de on behalf of David.Day@duke-software.com> wrote: > > > > > > I googled websockets, and sure enough, the RFC talks about the HTTP > > > protocol. I'm not going to try to code that. > > > > > > I've been in this business long enough to know there are always multiple > > > ways to get something done. I've got a lot of time invested in code > > > that creates both schema and xml for my data, and really don't want to > > > throw that time and effort away. > > > > > > Your suggestion earlier of having a piece of code running on a desktop > > > external to Basex to provide the communication link probably would work, > > > but it is starting to look kind of like a kluge now, I think. > > > > > > The scenario. > > > > > > Desktop code accepts the connection from the mainframe, and > > > connects to the Basex server. > > > > > > Would have to do a poll of Basex to determine if a new request had > > > been added to a request database. > > > > > > User, at a browser, would add a request to the database, and then > > > wait for the response. > > > > > > Code doing the poll would get the request and stuff it down the > > > pipe to the mainframe, then go back to its poll. Needs to be able to > > > recognize when data comes in on the connection as well. > > > > > > Response comes in, and the Basex request database is updated. > > > > > > Basex server sends the response to the browser. > > > > > > Would this work? > > > > > > > > > On 10/19/2017 3:08 PM, Kendall Shaw wrote: > > > > It does: HTTP. > > > > > > > > On 10/19/17, 12:42 PM, "basex-talk-bounces@mailman.uni-konstanz.de on behalf of Dave Day" <basex-talk-bounces@mailman.uni-konstanz.de on behalf of David.Day@duke-software.com> wrote: > > > > > > > > Unless websockets has some protocol sitting on top of standard TCP/IP > > > > communication, there shouldn't be an issue. > > > > > > > > > > > > > > > > On 10/19/2017 2:29 PM, Kendall Shaw wrote: > > > > > On 10/19/17, 10:54 AM, "Christian Grün" <christian.gruen@gmail.com> wrote: > > > > > > > > > > @Kendall: > > > > > > > > > > > If I understood, it was connecting from mainframe to the BaseX server over a socket that was wanted (I assume that means TCP). I think that websockets don’t connect to TCP sockets, without HTTP on both sides to establish the connection. So, if using HTTP would require implementing HTTP (or parts of it) in assembly language, that is not trivial, possibly. > > > > > > > > > > Yes, exactly. > > > > > > > > > > > > > > > @Marco: > > > > > > > > > > > not only browser clients but all other application (written in Java, nodeJS, Python ...) supporting websocket client libraries I suppose! ;-) > > > > > > > > > > Quite definitely! I’ll knock at your door once the first beta versions > > > > > will be around. If you believe you can give some helpful input on the > > > > > feature design in the GitHub issue, your feedback will be welcome. > > > > > > > > > > Great, in general. But, if one side is on the mainframe unless there is a websocket library there the problem would be the same. > > > > > > > > > > Kendall > > > > > > > > > > > On 10/19/17, 10:15 AM, "basex-talk-bounces@mailman.uni-konstanz.de on behalf of Marco Lettere" <basex-talk-bounces@mailman.uni-konstanz.de on behalf of m.lettere@gmail.com> wrote: > > > > > > > > > > > > Hi, > > > > > > not only browser clients but all other application (written in Java, > > > > > > nodeJS, Python ...) supporting websocket client libraries I suppose! ;-) > > > > > > Waiting eagerly for it ... > > > > > > M. > > > > > > > > > > > > On 19/10/2017 19:08, Christian Grün wrote: > > > > > > > Hi Dave, > > > > > > > > > > > > > > Kendall has already given you all relevant information (thanks!). > > > > > > > > > > > > > > Very early client bindings of BaseX provided support for bidirectional > > > > > > > connections. As the functionality was rarely used, and as only a few > > > > > > > bindings supported this part of the protocol, it was eventually > > > > > > > dropped. If you scan the history of our Wiki and the Github sources, > > > > > > > you will probably stumble upon the code snippets. > > > > > > > > > > > > > > The major reason is that most of today’s applications that are > > > > > > > interactive and are built on top of BaseX rely on HTTP. As soon as > > > > > > > WebSockets will be available [1], it will be possible to send data to > > > > > > > browser clients. > > > > > > > > > > > > > > Hope this helps, > > > > > > > Christian > > > > > > > > > > > > > > [1] https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_BaseXdb_basex_issues_1449&d=DwIDaQ&c=DS6PUFBBr_KiLo7Sjt3ljp5jaW5k2i9ijVXllEdOozc&r=JgwnBEpN1c-DDmq-Up2QMq9rrGyfWK0KtSpT7dxRglA&m=HTzHEkqTVFJxI8dfoxQV2nh-Sx_PWJdIVZl7i7EiaAM&s=EE2LEkklhkn_5oQHObR67dH0aMbWgLwEP9Zcx5JHmYM&e= > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Thu, Oct 19, 2017 at 7:01 PM, Dave Day <David.Day@duke-software.com> wrote: > > > > > > >> Hi Kendall, > > > > > > >> > > > > > > >> Thanks for the response. > > > > > > >> > > > > > > >> Seems to me there ought to be some way to get a Basex server to write a > > > > > > >> data stream to a socket other than as the result of running a query or a > > > > > > >> command. > > > > > > >> > > > > > > >> Maybe someone from the Basex developers group will chime in here. > > > > > > >> > > > > > > >> -- Dave > > > > > > >> > > > > > > >> > > > > > > >> > > > > > > >> On 10/18/2017 7:56 PM, Kendall Shaw wrote: > > > > > > >>> Someone can probably give a better answer. As far as I know there are no > > > > > > >>> xquery socket io functions provided with basex. You can write java to do > > > > > > >>> that or you can use an external process that communicates with the mainframe > > > > > > >>> and invoke that process from xquery, among other solutions. If you write > > > > > > >>> java you could look at the basex source to imitate how it uses sockets > > > > > > >>> and/or read about socket io in java. With that you could poll or send and > > > > > > >>> receive data to from the mainframe. > > > > > > >>> > > > > > > >>> To know if the mainframe connection is connected, I think you can use the > > > > > > >>> admin:sessions() function to see if it is connected or your program that > > > > > > >>> does the io could indicate that it is not connected if it can’t communicate. > > > > > > >>> > > > > > > >>> If you use xquery and RESTXQ to write the web application, and the same > > > > > > >>> basex instance is communicating with the mainframe then the xquery in your > > > > > > >>> application is already running in the basex server. > > > > > > >>> > > > > > > >>> For the second task it’s the same idea. You will need to read about how > > > > > > >>> updating works: > > > > > > >>> > > > > > > >>> https://urldefense.proofpoint.com/v2/url?u=http-3A__docs.basex.org_wiki_XQuery-5FUpdate&d=DwIDaQ&c=DS6PUFBBr_KiLo7Sjt3ljp5jaW5k2i9ijVXllEdOozc&r=JgwnBEpN1c-DDmq-Up2QMq9rrGyfWK0KtSpT7dxRglA&m=HTzHEkqTVFJxI8dfoxQV2nh-Sx_PWJdIVZl7i7EiaAM&s=_pSKI0JdH9q6EQwFZt5180gq24TDNk9RHCCCU9529v4&e= > > > > > > >>> > > > > > > >>> Or, reading this would tell you a lot: > > > > > > >>> > > > > > > >>> https://urldefense.proofpoint.com/v2/url?u=http-3A__docs.basex.org_wiki_Table-5Fof-5FContents&d=DwIDaQ&c=DS6PUFBBr_KiLo7Sjt3ljp5jaW5k2i9ijVXllEdOozc&r=JgwnBEpN1c-DDmq-Up2QMq9rrGyfWK0KtSpT7dxRglA&m=HTzHEkqTVFJxI8dfoxQV2nh-Sx_PWJdIVZl7i7EiaAM&s=d1UB9tyc9LKqvqvzoP9KPWE9x9Q2Va9Ge5qSYrTsEvA&e= > > > > > > >>> > > > > > > >>> Kendall > > > > > > >>> > > > > > > >>> On 10/18/17, 3:22 PM, "basex-talk-bounces@mailman.uni-konstanz.de on > > > > > > >>> behalf of Dave Day" <basex-talk-bounces@mailman.uni-konstanz.de on behalf of > > > > > > >>> David.Day@duke-software.com> wrote: > > > > > > >>> > > > > > > >>> Greetings list, > > > > > > >>> I have an idea of how I want to use Basex, and would like > > > > > > >>> to know > > > > > > >>> if what I am wanting to do will work. > > > > > > >>> My background is IBM mainframe, so the terminology I use is > > > > > > >>> influenced by that. > > > > > > >>> I have a server running in an address space on a mainframe. > > > > > > >>> Within > > > > > > >>> that address space are multiple tasks active. I believe the term > > > > > > >>> task > > > > > > >>> is equivalent to a thread in the distributed world. > > > > > > >>> I would like to have a task initialize when my server comes > > > > > > >>> up and > > > > > > >>> establish a connection to an instance of a Basex server. This task > > > > > > >>> would > > > > > > >>> be responsible for processing requests that come in from the Basex > > > > > > >>> server. > > > > > > >>> On a browser, I want to have a user connect to an instance > > > > > > >>> of the > > > > > > >>> Basex HTTP server. From the manual, it states that if the HTTP > > > > > > >>> server > > > > > > >>> is started, the client server is started as well. I don't know if > > > > > > >>> this > > > > > > >>> is all one instance of a Basex, or if there are actually two > > > > > > >>> processes > > > > > > >>> that get established. The HTTP server will need to know if the > > > > > > >>> connection has been established between the mainframe and the client > > > > > > >>> server. Is this possible, and if so, how would that be accomplished? > > > > > > >>> Assuming that the user on the browser needs some > > > > > > >>> information from > > > > > > >>> my mainframe server, is there a mechanism in place to facilitate the > > > > > > >>> HTTP server sending the request to the client server, and the client > > > > > > >>> server sending the request to the mainframe. Would the mainframe > > > > > > >>> have > > > > > > >>> to do a poll of some kind to see if there was data, or can the client > > > > > > >>> server push the request out across the connection? > > > > > > >>> If the above works, that is, Basex acting as a message > > > > > > >>> broker > > > > > > >>> between a browser and the mainframe, I want to be able to create > > > > > > >>> another > > > > > > >>> connection from another task on the mainframe to the Basex client > > > > > > >>> server > > > > > > >>> as a result of a user request on a browser, have that 2nd task create > > > > > > >>> a > > > > > > >>> new database, and start pushing xml across that 2nd connection. > > > > > > >>> While > > > > > > >>> that new database is being updated from the mainframe, can the user > > > > > > >>> on > > > > > > >>> the browser query the new database to get the most current data? > > > > > > >>> Regards, > > > > > > >>> -- Dave Day > > > > > > >>> > > > > > > >> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >