Re: Showstopper desktop runtime issue

On Thu, Mar 22, 2018 at 10:19 AM, Neel Patel <neel(dot)patel(at)enterprisedb(dot)com>
wrote:

> Hi,
>
> On Thu, Mar 22, 2018 at 3:19 PM, Dave Page <dpage(at)pgadmin(dot)org> wrote:
>
>> All,
>>
>> As you know, the 3.0 release is currently on hold as we discovered late
>> yesterday that the re-vamped desktop runtime will not run on Gnome 3.26 and
>> later. This is because the GTK project, and later Gnome, have removed
>> support for the System Tray on which the new runtime relies.
>>
>> They have replaced it with a notification mechanism, however this doesn't
>> really meet our needs as what we want is a place (the tray icon) to attach
>> a menu to control the pgAdmin server; we don't really use notifications as
>> such.
>>
>> I see a number of possible ways around this:
>>
>> 1) Return to the previous runtime. I think this is at best a short-term
>> solution, as the re-visited Annulen version of the QtWebKit seems to be
>> getting little attention at the moment, and this would re-introduce many
>> known bugs caused by WebKit.
>>
>> 2) Re-work the current runtime code to remove the tray icon, and utilise
>> desktop/start menu items to signal the running instance to show the logs,
>> configure the server, exit etc. This should work, but will be kinda klunky.
>>
>> 3) Put effort into polishing Joao's Electron based runtime. This might be
>> a good long term solution as it would remove the need to have any C++ code
>> of our own, and might allow us to use Electron's update mechanism to do
>> software updates. The downsides are that we would lose support for dockable
>> tabs (new windows only), and it wouldn't work on CentOS/RHEL 6 which we
>> currently support.
>>
>> Thoughts and comments please folks? How do we want to proceed? I'm
>> currently leaning towards 2 for v3, and possibly moving to 3 in the long
>> term.
>>
>
> As per our previous efforts on Qt's web rendering framework, I am not in
> favor of adding it again as it add lots of dependency and complexity.
> I would suggest to go for option 2 - To re-work the current runtime code
> and find out how to get already running instance of pgadmin4.
>

We have a shared memory interlock as well as temp/lock files that encode
the executable path into their name that handles that already, though it
doesn't seem to be working quite as expected when you run from a Mac
Appbundle (which Akshay is working on). The issue would be how we signal
the running instance and tell it to do something, e.g. exit.

--
Dave Page
Blog: http://pgsnake.blogspot.com
Twitter: @pgsnake

EnterpriseDB UK: http://www.enterprisedb.com
The Enterprise PostgreSQL Company