On Wed, 2014-03-05 at 00:00 +0100, Albert Cervera i Areny wrote:
> 2014-03-04 23:49 GMT+01:00 Mark Hayden (local) <[email protected]>: > > There is already an Android client for Tryton that I've used, but I find the > > idea of a platform-agnostic mobile solution appealing, so I will jump in... > > > > > > On Tue, 2014-03-04 at 12:30 -0800, Emma wrote: > > > > The goal of this method is to have an exactly the same behavior between > > the tryton and sao. > > > > > > If that is a stated goal (which makes sense for general desktop use I think) > > then there could be a lot of friction in terms of accommodating mobile. > > Making a full featured, richly functional app that is a drop in replacement > > for a native desktop app conflicts in many ways with mobile-friendly > > responsive design. The native client is not "responsive" so making SAO > > responsive conflicts with the goal of "native look and feel" > > > > > > > > Well, with this patch, the behavior of sao stays EXACTLY the same (it's just > > a poc so far, of course it needs tweaking) as the desktop until you reach > > the trigger minimum sizes. > > > > In the poc: > > > > - less than 992px wide: single column > > > > - less than 768px wide: menu disapears (A toggle button would be needed) > > > > > > > > It useful to users > > to see some information from mobile > > > > > > I think this is an interesting "hack" or proof-of-concept as you say, but > > not the approach I would take. Firstly screen resolution is not a reliable > > way to determine the nature of a device on its own; my mobile has only as > > 12cm screen but the resolution is full HD (1920x1080) and so you would have > > to query for other information (pixel density, physical size...). > > > > Second I think the needs of the desktop and those of mobile diverge enough > > that a specific interface for mobile/responsive environment is a better > > solution. > > > > > > > > So since it's usefull for users on mobile, why not make it comfortable as > > well? > > > > This patch so far doesn't change any logic on the js side, it just replaces > > a table with div's (colspan implemented but not xexpand nor xfill) > > > > I think using the "engine" of sao to provide an alternative interface can > > minimise duplication, but I think the best approach (though more work) is to > > make a set of templates/front end from scratch in a "mobile-first" design. > > Doing client-sniffing and then applying mixins, substitutions, etc. to the > > full client interface would result in a sub-optimal experience and make > > maintenance troublesome. > > > > So my preference would be to use the sao architecture but do a "clean sheet" > > responsive client (perhaps called "sao-lite"). Note I say "Responsive" and > > not "mobile" because I think such a client would have its uses on both > > mobile and non-mobile applications (kiosks, point-of-sale, front-line > > employees, etc) where the full layout and content of the Tryton client are > > not required. > > > > Mark, I think you make very interesting considerations. I know very > little about web development, but maybe using bootstrap could be a > step in the right direction even if we eventually see it is better to > have a sao-lite version, because both sao and sao-lite would share > most of the framework. > Thanks for the feedback. I am a bit of a novice in the use of these frameworks as well, but I have had a chance to play with both bootstrap and Foundation 5. So far I prefer Foundation to bootstrap as it is a bit more "lightweight" and was optimized for mobile-first (though it works quite well on responsive websites in general). It has been mentioned there is a "PDA mode" for the native/GTK client, so the idea of having the full sao and an "sao-lite" mode of operation, with the latter having a separate presenation layer on top of the same overall sao architecture, would mirror that approach and the user could explicitly pick which sao client to visit without us guessing the choice for them (perhaps a high-res tablet with a bluetooth keyboard would work for a full client, and maybe a casual user on a PC would still like to use the lite client). Also there would be less need to concern ourselves with interfering with the performance or functionality of the main (full) client. Just my thoughts...
