PDA

View Full Version : DataFlex OAuth 2 component beta 2



Mike Peat
23-Sep-2015, 10:10 AM
I am uploading a second beta version of the DataFlex OAuth 2.0 component workspace to correct some minor errors that have appeared during development of the Office 365 API component.

There is very little changed:


At Harm's advice (and because I found that otherwise JavaScript Booleans were being returned as strings) I have explicitly added all of the DataFlex web properties as proper framework properties in the JavaScript
At Marcia's suggestion, the sample application should now "trim" the operation string before use
Because Roel wanted to do it and it was the easiest way for me to show him how, a "Meetup.com" sample has been added
Fixed (I hope) are some minor problems which had emerged with dates (when calculating the amount of time left in an access grant): these would only affect you if you were (a) using a non-US date format and (b) ran the program initially under the debugger, doing your login to whichever provider there, then exited the debugger, but continued to run the application. In that case, on certain days of certain months, the date/time an access token was granted could appear to be later than the current date/time, causing a calculation error (I am guessing that this is because the debugger uses US dates, not the local date format)


9277

Updated documentation is here (https://docs.google.com/document/d/14Kvk0C-vZXBrORh6IjlEzk63Z1KuYHQ3vL5CzpXHXmY/edit?usp=sharing).

Mike

Samuel Pizarro
26-Oct-2020, 05:55 PM
Hi Mike

Is this the latest version of this component/lib ?

Mike Peat
27-Oct-2020, 03:45 AM
Samuel

Yes.

Mike

Samuel Pizarro
27-Oct-2020, 07:48 AM
Thanks!

DaveR
28-Oct-2020, 08:13 AM
...

.... In that case, on certain days of certain months, the date/time an access token was granted could appear to be later than the current date/time, causing a calculation error (I am guessing that this is because the debugger uses US dates, not the local date format)


...
Mike

That last one... :rolleyes:

I've recently cracked an 18-months-old, applies-to-one-person-only bug which was as obscure as that. Turns out 1 user (we think - but we have some furloughed) of the 80-100 always exited from a given function/dialog with <escape> rather than one of the buttons. Turns out that this is the only dialog where I hadn't sat on escape. Turns out that escaping this dialog meant that a couple of entries in the program main data struct went unfilled. Turns out that, in combination with other circumstances, those empty struct members had a knock-on effect in what was written to the main transaction file. And finally turns out that this unusual value in the transaction file blows up a report that is run unsupervised and sends its result direct to a vendor. Oh, and the vendor never mentioned it.

Oh, and this being IT, all those problems had to be solved in reverse order. Except for identifying the cashier, which happened by accident "I don't like the new version".