If you drive a RS232 line by an output of 74HC125 gate instead of special RS232 driver, then you loose two important properties, which are necessary to allow reliable signal transmission:

1. RS232 signal level standard is violated, due to the lack of negative signal voltage (<0V!) for 'Mark', when using 74HC125.

2. Normally, slew rate is heavily limited when using a standard RS232 driver, but when using 74HC125 slew rate is rather unlimited, resulting in possible EMI trouble (capacitive cross coupling between the lines, mistriggering due reflections caused by improper line termination, etc.)!

RS232 standard does not use line termination, means ends of cable are not terminated by impedances matching the characterristic impedance of cable (transmission line). The only way to prevent the disadvantages of mismatching (reflections, for instance) then is to use slew rate limited line drivers. So, it's not a good idea to use 74HCMOS gate output instead of a RS232 line driver!

The consequence for your application is to use TWO different MAX232, each with their output directly connected to the according connector. The switching between both MAX232 should be done at the INPUTS of the two MAX232 then. If you insist on furtherly using these 74HC125 gates, then connect them between the micro's output and the according input of MAX232. When an 74HC125 gate is disabled, then its output goes tristate and the internal pull-up of MAX232 of about 400kOhm makes the output of MAX232 going low, emitting a negative voltage to the line, which represents idle level (Mark).
If the 74HC125 is enabled, on the other hand, then micro has full control over the according RS232 line.

In some cases internal 400kOhm pull-ups might be a bit too high, possibly resulting in increased susceptibility against noise at these pins. In such a case you could add 10kOhm pull-ups at these inputs.

Kai