Update strategy?

[Graham Samuel]

Apologies if this has come up relatively recently, but I have not been very attentive to the list for a bit…

I have a desktop app (though in principle it could be on mobile) which uses a variant of the ‘splashscreen’ structure. What happens is that the app as seen by the operating system is actually an initialisation stack, which then calls in a stack containing the bulk of the script and graphics for the app and executes that. (I call this a ‘data stack’ although this is a bit of misnomer, as it does contain the script libraries that do most of the work.) The clean (template) copy if this data stack is stored in the app’s resources folder, and is loaded the first time the app is started; thereafter the user can alter the data stack, and the altered version is saved in the application data folder. There is a reset facility for going back to the clean template.

When a new version of the app is installed, the splash stack detects that the data stack is in old format (actually, that it has an old version number) and forces a reset, thus ensuring that the latest data stack comes into use.

All this works quite nicely, but I notice so many apps that automatically check for updates, providing a dialog to the user offering to do the update: if the user agrees, then the update takes place without further intervention.

I can kind of see how to do this (the splash stack checks with the server where the app originated to see if there is a more up to date version, then somehow replaces itself), but are there any gotchas in this approach? One I can think of so far is when the user runs the app offline, so that any approach to the server will fail - not sure how to detect that. Also, so far I am vague about how a running standalone can replace itself - something do do with file names, perhaps?

I’d be grateful for any advice or experience.

Graham