On 30-9-2014 2:48, Josh Andler wrote:
On Mon, Sep 29, 2014 at 4:40 PM, Johan Engelen <jbc.engelen@...2592...> wrote:
On 25-9-2014 17:01, Martin Owens wrote:
Of course the miserated speed of inkscape on anything other than a core2 is a real big problem. But something I don't think that can be tackled until we have paid developers who do nothing but speed improvement bugs.
I do not think we need paid developers to look into this. We just need *any* developer to work on this. We don't have to render icons from SVG upon startup, load all the systems fonts, etc.
Well, I'd be interested to know what experimental starting up on a lower end machine is like. It may not be a magical and huge difference, but Liam made some great improvements with launch speed there.
Tbh, on my core-i7, the startup time is long enough to warrant a splash screen. Plus, what is the downside of a splash screen? :)
As for the icons, if I recall correctly, they get cached on first launch and only get re-rendered if the source file gets modified.
I count 1044 calls to SPPath::build upon starting trunk (after starting it several times). Perhaps it's not the icons, but something is creating a lot of SVG objects. Otoh, the boottime is dominated by something else, because the majority of time is spent before the calls to SPPath::build are performed; the part /with/ calls to SPPath::build appears fast on my machine.
-Johan