On Sat, 2006-09-09 at 09:15 -0400, Adam Hooper wrote: > On Sat, 2006-09-09 at 00:36 -0400, John Moser wrote: > > On Fri, 2006-09-08 at 23:48 -0400, Adam Hooper wrote: > > > > So, technically, there's no reason not to do it. Realistically, it's a > > > matter of putting in a lot of effort. > > > > Technically there's almost no reason to do it as well, am I correct? > > (besides, of course, the convenient smoke-and-mirrors ideas of "some > > users prefer X to Y" and "sometimes X has bugs not in Y and vice versa") > > Well, there's certainly reason to do it -- otherwise, you wouldn't have > written to the list in the first place :). A small part of the reason is > purely political -- "look, Mozilla, we don't need you, we *choose* you, > so don't give us reason to switch". With the existing proof-of-concept, I move that Mozilla is not Microsoft and so does not care, they'll just push Firefox anyway. As long as IE dies they're happy. > this is already available to us. The rest is technical: smaller memory > footprint, potentially faster page loads, works better for a few web > pages, etc. > Are you saying it DOES have a smaller footprint, or that it possibly does? > But webcore has a large set of bugs. Having used Gecko for ages, > Epiphany developers are aware of most of the icky Mozilla bugs, but the > webcore ones would be brand new. That leads to: > Webcore has been unmaintained for 22 months according to CVS. > - a huge amount of effort to work around some annoyances (witness just > how much work went into Epiphany's "focus location bar on new tab" > behaviour) > - a huge amount of effort to fix webcore where it doesn't play nice with > Epiphany > - after all that, double the current effort just to maintain Epiphany > Possibly, could Webcore be taught to behave in a minimalist manner (i.e. put the JavaScript engine into another module, separate out whatever else is possible) so that it could be loaded in various configurations, including: - A complete Web rendering engine - A minimalized, low footprint rendering engine This could possibly share the 'double' effort by paving the way for GtkHtml 4.0, if you could get the memory footprint low enough. I'm not sure what GtkHtml is based on right now but it's only used for minimal things, no JavaScript, etc; and this means I'm not sure if you could get something like Webcore matching its low low memory footprint in a minimal configuration. Anyway, my point is Webcore could possibly be useful in places other than simply as an Epiphany back-end. Evolution could use it for HTML desktop for example. OK so I have no good examples ;) Another consideration, if the back-end plug architecture is good enough, the 'old' Webcore/Mozilla modules should work with 'new' Epiphany after upgrades, like XMMS plug-ins work with XMMS after upgrade. > So, the cost is high. We're not quite sure just how much benefit there > is. Who knows -- there's a chance that the benefit outweighs the cost > (at least to certain people). And the beauty of open-source is that > anybody who's willing can go ahead and do it, and then we'll really > know. > > But nobody wants to :). Because the cost is too high, and the benefit is not visible :) Anyway, thanks. At this point I know that this is a cost-benefit issue and not a political issue (you wouldn't believe how vehemently some projects fight for the status quo). > -- John Moser <john r moser gmail com>
Attachment:
signature.asc
Description: This is a digitally signed message part