On Sun, Feb 5, 2012 at 1:10 PM, Phil Charlesworth <[email protected]> wrote: > On 04/02/12 20:04, lkcl luke wrote: >> On Sat, Feb 4, 2012 at 7:38 PM, John P Charlesworth >> <[email protected]> wrote: >> >>> >>> On Sat, Feb 4, 2012 at 5:32 PM, lkcl luke<[email protected]> wrote: >>> >>>> On Sat, Feb 4, 2012 at 5:23 PM, Phil Charlesworth >>>> <[email protected]> wrote: >>>> >>>>> On 04/02/12 14:37, lkcl luke wrote: >>>>> >>>>>> On Sat, Feb 4, 2012 at 2:29 PM, Phil Charlesworth >>>>>> <[email protected]> wrote: >>>>>> >>>>>> >>>>>> >>>>>>>> the thing is that i'm reluctant to change something as fundamental >>>>>>>> as >>>>>>>> DOM.setAttribute without analysing its full impact across *all* the >>>>>>>> pyjd engines. >>>>>>>> >>>>>>>> anything like this is going to require comprehensive and thorough >>>>>>>> testing. >>>>>>>> >>>>>>>> l. >>>>>>>> >>>>>>>> >>>>>>> Fair enough P. >>>>>>> >>>>>>> >>>>>> it's a bugger, innit? :) >>>>>> >>>>>> would anyone like to give this a shot? modify the code back to not >>>>>> have the "str" conversion, and see what happens? >>>>>> >>>>>> we need to find out what breaks - if anything - and/or what the >>>>>> implications are. >>>>>> >>>>>> l. >>>>>> >>>>> Tested KitchenSink on Windows 7 and got same error messages (from >>>>> getText()) when typing pound chars into the TextBox widgets on the Text >>>>> page. Those errors certainly go away when the str() calls are removed. >>>>> >>>>> However, something very interesting - I now have pyjamas running on >>>>> Debian in a VirtualBox - and KitchenSink there shows no sign of this >>>>> problem! It is using hulahop+webview+Webview it says. >>>>> So perhaps, after all, it IS something specific to MSHTML. >>>>> Phil >>>>> >>>> hmmm, then what could be done there is to put in a platform-override >>>> in .. mmm.... library/pyjamas/platform/DOMmshtml.py which *doesn't* >>>> have the str call >>>> >>>> that would do the trick. >>>> >>> >>> OK, I will do that. It will need an override for TextBoxBase and for DOM but >>> the necessary files already exist, so it's just a case of editing them. >>> I'll put a comment and a diff file on the Issue when I've done it. >>> >> star. >> >> btw, save off committing for a while, i'm just dealing with the >> platform overrides, something i've been meaning to do for a looong >> time. >> >> l. > Comment and diff file added to Issue #681 > http://code.google.com/p/pyjamas/issues/detail?id=681
ack, good stuff phil. i simplified the getAttribute function - there's no need to test "if is None return None" :) actually it could be simplified further with "getAttribute = getattr" but i thought it best to leave it as an explicit function call, esp. in a platform override. > Incidentally, the new system for platform overrides is a vast > improvement. Thanks for doing that. achh, no problem. slightly confused as to why i didn't do it earlier. i think it will make things much clearer now that there even _is_ an override system: putting the files in separate subdirectories made it non-obvious. l.
