For Review: Change 20070403-maxcarlson-s Summary: Fix DHTML proxied loads in DHTML for

Change 20070403-maxcarlson-s by maxcarlsonplastik.lan on 2007-04-03  
06:40:38 PDT
     in /Users/maxcarlson/openlaszlo/legals-clean
     for 
http://svn.openlaszlo.org/openlaszlo/branches/legals

Summary: Fix DHTML proxied loads in DHTML for apps loaded
from urls  
beginning with any protocol

New Features:

Bugs Fixed:

Technical Reviewer: promanik
QA Reviewer: ptw
Doc Reviewer: (pending)

Documentation:

Release Notes:

Details: LzBrowser.js - Leave urls beginning with slash or
with a  
colon before the slash alone, 'specially for Mr. Pedantic



Tests:


Files:
M      WEB-INF/lps/lfc/services/platform/dhtml/LzBrowser.js

Changeset: http://svn.openlaszlo.org/openlaszlo/patches/20070403-

maxcarlson-s.tar

Mr. Pedantic sez:

  if (slash == 0 || (colon > 0 && (slash == -1 ||
colon < slash)) {

But wait, then there is:

   http:foo

which is a relative url with an explicit scheme.  Shouldn't
that also  
prepend the load directory?  It seems you really have to
parse the  
url and then decide whether the path of that url is relative
or not.   
Don't we already have an url parser utility?

On 2007-04-03, at 09:42 EDT, Max Carlson wrote:

> Change 20070403-maxcarlson-s by maxcarlsonplastik.lan on  
> 2007-04-03 06:40:38 PDT
>     in /Users/maxcarlson/openlaszlo/legals-clean
>     for 
http://svn.openlaszlo.org/openlaszlo/branches/legals
>
> Summary: Fix DHTML proxied loads in DHTML for apps
loaded from urls  
> beginning with any protocol
>
> New Features:
>
> Bugs Fixed:
>
> Technical Reviewer: promanik
> QA Reviewer: ptw
> Doc Reviewer: (pending)
>
> Documentation:
>
> Release Notes:
>
> Details: LzBrowser.js - Leave urls beginning with slash
or with a  
> colon before the slash alone, 'specially for Mr.
Pedantic 
>
>
> Tests:
>
>
> Files:
> M     
WEB-INF/lps/lfc/services/platform/dhtml/LzBrowser.js
>
> Changeset: http://svn.openlaszlo.org/openlaszlo/patches/20070403-

> maxcarlson-s.tar

(Late pile on; my apologies.)

In my experience, you always lose in the end if you try to
pick apart  
a URL without using a general URL parsing routine. You are
just  
engineering bugs unless you write comprehensive unit tests
every time  
you manually pick apart of URL, and what would be the sense
of that?

jim

On Apr 3, 2007, at 7:11 AM, P T Withington wrote:

> Mr. Pedantic sez:
>
>  if (slash == 0 || (colon > 0 && (slash ==
-1 || colon < slash)) {
>
> But wait, then there is:
>
>   http:foo
>
> which is a relative url with an explicit scheme. 
Shouldn't that  
> also prepend the load directory?  It seems you really
have to parse  
> the url and then decide whether the path of that url is
relative or  
> not.  Don't we already have an url parser utility?
>
> On 2007-04-03, at 09:42 EDT, Max Carlson wrote:
>
>> Change 20070403-maxcarlson-s by maxcarlsonplastik.lan on  
>> 2007-04-03 06:40:38 PDT
>>     in /Users/maxcarlson/openlaszlo/legals-clean
>>     for 
http://svn.openlaszlo.org/openlaszlo/branches/legals
>>
>> Summary: Fix DHTML proxied loads in DHTML for apps
loaded from  
>> urls beginning with any protocol
>>
>> New Features:
>>
>> Bugs Fixed:
>>
>> Technical Reviewer: promanik
>> QA Reviewer: ptw
>> Doc Reviewer: (pending)
>>
>> Documentation:
>>
>> Release Notes:
>>
>> Details: LzBrowser.js - Leave urls beginning with
slash or with a  
>> colon before the slash alone, 'specially for Mr.
Pedantic 
>>
>>
>> Tests:
>>
>>
>> Files:
>> M     
WEB-INF/lps/lfc/services/platform/dhtml/LzBrowser.js
>>
>> Changeset: http://svn.openlaszlo.org/openlaszlo/patches/20070403-

>> maxcarlson-s.tar
>

I agree.  We should have a task to create a general URL
parse/merge  
utility.  I think we have the beginnings of it already, but
it needs  
to be teased apart.  If I am not mistaken, in this case Max
is trying  
to create the base URL that will be used as the default for
missing  
bits when the general URL parser is invoked, so there is a
little bit  
of chicken/egg problem.  But I think we could solve that
with some  
remodularization.  We want a Java-like facility where you
can parse  
an URL and supply a second URL as the one to take defaults
from.

On 2007-04-09, at 23:49 EDT, Jim Grandy wrote:

> (Late pile on; my apologies.)
>
> In my experience, you always lose in the end if you try
to pick  
> apart a URL without using a general URL parsing
routine. You are  
> just engineering bugs unless you write comprehensive
unit tests  
> every time you manually pick apart of URL, and what
would be the  
> sense of that?
>
> jim
>
> On Apr 3, 2007, at 7:11 AM, P T Withington wrote:
>
>> Mr. Pedantic sez:
>>
>>  if (slash == 0 || (colon > 0 && (slash
== -1 || colon < slash)) {
>>
>> But wait, then there is:
>>
>>   http:foo
>>
>> which is a relative url with an explicit scheme. 
Shouldn't that  
>> also prepend the load directory?  It seems you
really have to  
>> parse the url and then decide whether the path of
that url is  
>> relative or not.  Don't we already have an url
parser utility?
>>
>> On 2007-04-03, at 09:42 EDT, Max Carlson wrote:
>>
>>> Change 20070403-maxcarlson-s by maxcarlsonplastik.lan on  
>>> 2007-04-03 06:40:38 PDT
>>>     in
/Users/maxcarlson/openlaszlo/legals-clean
>>>     for 
http://svn.openlaszlo.org/openlaszlo/branches/legals
>>>
>>> Summary: Fix DHTML proxied loads in DHTML for
apps loaded from  
>>> urls beginning with any protocol
>>>
>>> New Features:
>>>
>>> Bugs Fixed:
>>>
>>> Technical Reviewer: promanik
>>> QA Reviewer: ptw
>>> Doc Reviewer: (pending)
>>>
>>> Documentation:
>>>
>>> Release Notes:
>>>
>>> Details: LzBrowser.js - Leave urls beginning
with slash or with a  
>>> colon before the slash alone, 'specially for
Mr. Pedantic 
>>>
>>>
>>> Tests:
>>>
>>>
>>> Files:
>>> M     
WEB-INF/lps/lfc/services/platform/dhtml/LzBrowser.js
>>>
>>> Changeset: http://svn.openlaszlo.org/openlaszlo/patches/20070403-

>>> maxcarlson-s.tar
>>
>

Yep, I had this sick feeling the whole time I was fixing the
bug.  I do 
want to remodularize, but I also wanted to get lzunit
passing again. 
Here's a bug to track this: http:
//jira.openlaszlo.org/jira/browse/LPP-3864

-Max

P T Withington wrote:
> I agree.  We should have a task to create a general URL
parse/merge 
> utility.  I think we have the beginnings of it already,
but it needs to 
> be teased apart.  If I am not mistaken, in this case
Max is trying to 
> create the base URL that will be used as the default
for missing bits 
> when the general URL parser is invoked, so there is a
little bit of 
> chicken/egg problem.  But I think we could solve that
with some 
> remodularization.  We want a Java-like facility where
you can parse an 
> URL and supply a second URL as the one to take defaults
from.
> 
> On 2007-04-09, at 23:49 EDT, Jim Grandy wrote:
> 
>> (Late pile on; my apologies.)
>>
>> In my experience, you always lose in the end if you
try to pick apart 
>> a URL without using a general URL parsing routine.
You are just 
>> engineering bugs unless you write comprehensive
unit tests every time 
>> you manually pick apart of URL, and what would be
the sense of that?
>>
>> jim
>>
>> On Apr 3, 2007, at 7:11 AM, P T Withington wrote:
>>
>>> Mr. Pedantic sez:
>>>
>>>  if (slash == 0 || (colon > 0 &&
(slash == -1 || colon < slash)) {
>>>
>>> But wait, then there is:
>>>
>>>   http:foo
>>>
>>> which is a relative url with an explicit
scheme.  Shouldn't that also 
>>> prepend the load directory?  It seems you
really have to parse the 
>>> url and then decide whether the path of that
url is relative or not.  
>>> Don't we already have an url parser utility?
>>>
>>> On 2007-04-03, at 09:42 EDT, Max Carlson
wrote:
>>>
>>>> Change 20070403-maxcarlson-s by
maxcarlsonplastik.lan on 2007-04-03 
>>>> 06:40:38 PDT
>>>>     in
/Users/maxcarlson/openlaszlo/legals-clean
>>>>     for 
http://svn.openlaszlo.org/openlaszlo/branches/legals
>>>>
>>>> Summary: Fix DHTML proxied loads in DHTML
for apps loaded from urls 
>>>> beginning with any protocol
>>>>
>>>> New Features:
>>>>
>>>> Bugs Fixed:
>>>>
>>>> Technical Reviewer: promanik
>>>> QA Reviewer: ptw
>>>> Doc Reviewer: (pending)
>>>>
>>>> Documentation:
>>>>
>>>> Release Notes:
>>>>
>>>> Details: LzBrowser.js - Leave urls
beginning with slash or with a 
>>>> colon before the slash alone, 'specially
for Mr. Pedantic 
>>>>
>>>>
>>>> Tests:
>>>>
>>>>
>>>> Files:
>>>> M     
WEB-INF/lps/lfc/services/platform/dhtml/LzBrowser.js
>>>>
>>>> Changeset: 
>>>> http://svn.openlaszlo.org/openlaszlo/patch
es/20070403-maxcarlson-s.tar
>>>
>>
> 

-- 
Regards,
Max Carlson
OpenLaszlo.org