Automatic front-end startup

The Genero can start a graphical front-end automatically, when the runtime system and the front-end reside on the same computer.

The GUI server autostart feature is supported for backward compatibility, and should be avoided. Consider using other solutions to start your Genero application, such as GDC shortcuts for desktop applications, the Genero Application Server (GAS) for web applications, or the "Run On Server" solution for mobile applications.

When a program starts in graphical mode, the runtime system tries to open a connection to the graphical front-end set by the FGLSERVER environment variable. This requires having the front-end already started and listening to the TCP port specified by FGLSERVER.

In some configurations, such as X11 workstations or METAFRAME/Citrix Winframe or Microsoft™ Windows® Terminal Server, each user may want to start his own front-end to have a dedicated process. This can be done by starting the front-end automatically when the program executes, based on settings in the DISPLAY (X11) or SESSIONNAME/CLIENTNAME (WTSE) environment variables.

Automatic front-end startup settings are defined with gui.server.autostart.* entries in FGLPROFILE. In these FGLPROFILE entries, the term "GUI server" refers to the graphical front end.

In a first time, the runtime system tries to establish the connection without starting the front-end (in a normal usage, it is already started). The front-end is identified by the FGLSERVER environment variable. If FGLSERVER is not defined, it defaults to localhost:0, except if gui.server.autostart.wsmap entries are defined in FGLPROFILE. When wsmap entries are defined, workstation id to GUI server id mapping takes place and FGLSERVER defaults to localhost:n, where n is the GUI server number found from wsmap entries.

If this first connection fails and the gui.server.autostart.cmd entry is defined, the runtime system executes the command to start the GUI server, then waits for n seconds as defined by gui.server.autostart.wait entry, and after this delay tries to connect to the front-end. If the connection fails, it tries again for a number of attempts defined by the gui.server.autostart.repeat entry. Finally, if the last attempt fails, the runtime system stops with a GUI connection error -6300.

If the gui.server.autostart.cmd entry is not defined, neither workstation id to GUI id mapping, nor automatic front-end startup is done.

Here is a detailed description of each gui.server.autostart FGLPROFILE entry:

The cmd entry is used to define the command to be executed to start the front-end:

gui.server.autostart.cmd = "/opt/app/gdc-2.30/bin/gdc -p %d -q -M"

Here, %d will be replaced by the TCP port the front-end must listen to.

By default the runtime system waits for two seconds before it tries to connect to the front-end. You can change this delay with the wait entry:

gui.server.autostart.wait = 5   -- wait five seconds

The runtime system tries to connect to the front-end ten times. You can change this with the repeat entry:

gui.server.autostart.repeat = 3 -- repeat three times

The following FGLPROFILE entries can be used to define workstation id to front-end id mapping:

gui.server.autostart.wsmap.max = 3
gui.server.autostart.wsmap.0.names = "fox:1.0,fox.sxb.4js.com:1.0"
gui.server.autostart.wsmap.1.names = "wolf:1.0,wolf.sxb.4js.com:1.0"
gui.server.autostart.wsmap.2.names = "wolf:2.0,wolf.sxb.4js.com:2.0"

The first wsmap.max entry defines the maximum number of front-end identifiers to look for. The wsmap.N.names entries define a mapping for each GUI server, where N is the front-end identifier. The value of those entries defines a comma-separated list of workstation names to match. If no wsmap entries are defined, the GUI server number will default to zero.

For gui.server.autostart.wsmap entries, the first GUI server number starts at zero.

On X11 configurations, a workstation is identified by the DISPLAY environment variable. In this example, fox:1.0 identifies a workstation that will make the runtime start a front-end with the number 1.

On Windows Terminal Server, the CLIENTNAME environment variable identifies the workstation. If no corresponding front-end id can be found in the wsmap entries, the front-end number defaults to the id of the session defined by the SESSIONNAME environment variable, plus one. The value of this variable has the form protocol#id; for example, RDP-Tcp#4 would automatically define a front-end id of 5 (4+1).

If the front-end processes are started on the same machine as the runtime system, you do not need to set the FGLSERVER environment variable. This will then default to localhost:id, where id will be detected from the wsmap workstation mapping entries.

If the front-end is executed on a middle-tier machine that is different from the application server, MIDHOST for example, you can set FGLSERVER to MIDHOST without a GUI server id. The workstation mapping will automatically find the id from the wsmap settings.