here's the current diff against trunk to remove those X's. :) diff -r 5d9a66496e58 gluon/http.py --- a/gluon/http.py Fri Feb 22 10:18:58 2013 -0600 +++ b/gluon/http.py Fri Feb 22 12:50:11 2013 -0800 @@ -97,9 +97,6 @@ if not body: body = status if isinstance(body, str): - if len(body) < 512 and \ - headers['Content-Type'].startswith('text/html'): - body += '<!-- %s //-->' % ('x' * 512) # trick IE headers['Content-Length'] = len(body) rheaders = [] for k, v in headers.iteritems():
cfh On Saturday, December 15, 2012 11:20:43 AM UTC-8, Massimo Di Pierro wrote: > > If no objection I will remove those. > > On Saturday, 15 December 2012 08:12:24 UTC-6, Niphlod wrote: >> >> nope. I'm +1 on all the line "remove those characters". >> My point was to eliminate that from http alltogether, or (but it will >> require more checks) remove that from restful AND @service calls. >> >> On Saturday, December 15, 2012 12:50:17 AM UTC+1, howesc wrote: >>> >>> is anybody using request.restful *and* needs the 512 bytes in a restful >>> response? i'm inclined to only skip those bytes for restful requests >>> (because they are usually not displayed by browsers). >>> >>> thanks, >>> >>> cfh >>> >>> On Friday, December 14, 2012 2:48:39 PM UTC-8, Niphlod wrote: >>>> >>>> Please... let be sure that those injected characters are going to be >>>> replied only to a browser request, possibly only IE. Technically as long >>>> as >>>> the gzipped body stays over 512 byte IE will show the page. >>>> >>>> Lets not forget, pleeeease, that the thread started requesting to >>>> delete those nasty 512 bytes (and I'm more and more inclined to forget >>>> about IE error pages): let's keep them 512 and not make them 90000 :-P >>>> >>>> On Friday, December 14, 2012 9:42:01 PM UTC+1, Massimo Di Pierro wrote: >>>>> >>>>> This is a problem. How about injecting more characters instead of >>>>> less. How about an image encoded in ascii? >>>>> >>>>> On Friday, 14 December 2012 13:44:24 UTC-6, Niphlod wrote: >>>>>> >>>>>> problem with older browser is : retrieve a working copy of it..... >>>>>> However, http://core.trac.wordpress.org/ticket/8942 and the >>>>>> following >>>>>> http://www.clintharris.net/2009/ie-512-byte-error-pages-and-wordpress/, >>>>>> http://support.microsoft.com/kb/294807 seems to point in the >>>>>> direction of < 7. >>>>>> Others sites include IE7 just referencing the "friendly error pages" >>>>>> item. >>>>>> >>>>>> Unfortunately it seems that in IE8 the problem persists (just >>>>>> checked): friendly error page kicks in. >>>>>> >>>>>> However, I'm saying: >>>>>> - the ticket page in rewrite.py is filled with characters already >>>>>> - we add them to the "temporarily down for maintenance" in main.py >>>>>> instead of injecting them on the HTTP() method, that can be (and its >>>>>> being) used also for interacting with non-browser clients. >>>>>> Cons: if anyone is doing >>>>>> raise HTTP(404, 'item not found') >>>>>> it won't display on IE. I'm positive though that if anyone is doing >>>>>> that an error "item not found" is not very much more informative than >>>>>> the >>>>>> "friendly page" of IE, and if it's needed "badly" a custom error page is >>>>>> prepared (and returned) >>>>>> >>>>>> PS: with gzip enabled it doesn't work anyway (meaning right now >>>>>> adding 512 "x" doesn't work). >>>>>> >>>>>> The only trick is resorting to HTTP(404, [something]) to skip the >>>>>> injecting feature.... >>>>>> >>>>>> -- --- You received this message because you are subscribed to the Google Groups "web2py-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to web2py+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.