Mailing-List: contact user-help@couchdb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@couchdb.apache.org
Received-SPF: pass (nike.apache.org: domain of bchesneau@gmail.com designates
 209.85.128.45 as permitted sender)
MIME-Version: 1.0
In-Reply-To: 
 <CA+UPnOX3xV7mqLvc_FtYJ711v44Gyjt8uDtG7rYUpYFhy-0x4w@mail.gmail.com>
References: 
 <CA+UPnOWc=X1Z+w3zXnnBWaLHo7zWq9igUEpkiaabac1vVCjnqw@mail.gmail.com>
	<CA+UPnOUYv6+66mV55w+J+57sgQivV5pvUTOhHfgf5m25LPzvUQ@mail.gmail.com>
	<CAL+Y1ntEXYcFDaazydD31j7cgdbeyRDEsJiFgOWK3Jao5LnVJg@mail.gmail.com>
	<CAPNuaCgdXXniOrgV1HxyUHpydafnGaB4L4rAOgR7qQNFrSxjLw@mail.gmail.com>
	<CAJNb-9q0s5p5pJO1b7wTtDGExcAGq_cG4veCKzwyLCJF5vit0w@mail.gmail.com>
	<CAPNuaCghrqhpOSLi20p_-zHFRZuVzHH5Hiuh43uRRyTOrLP1ew@mail.gmail.com>
	<CA+UPnOX3xV7mqLvc_FtYJ711v44Gyjt8uDtG7rYUpYFhy-0x4w@mail.gmail.com>
Date: Wed, 17 Apr 2013 10:08:11 +0200
Message-ID: 
 <CAJNb-9qDcJFUEm5qhWY2kE8Wwpn4VkzHq5BPLYuwwKnwDAU0Zg@mail.gmail.com>
Subject: Re: startkey endkey problem
From: Benoit Chesneau <bchesneau@gmail.com>
To: "user@couchdb.apache.org" <user@couchdb.apache.org>
Content-Type: text/plain; charset=ISO-8859-1

I mixed maills... sorry :)

On Wed, Apr 17, 2013 at 10:04 AM, Stanley Iriele
<siriele@breaktimestudios.com> wrote:
> Hopefully this helps someone somewhere... I'll make a tucket for this
> tomorrow... (Hopefully)
> On Apr 17, 2013 1:03 AM, "Stanley Iriele" <siriele2x3@gmail.com> wrote:
>
>> Hey....thanks for the clarity.. But that does not really apply to this
>> case....the way I'm requesting days fortunately doesn't bite me in the ass
>> :-) even though its string ordering... So..1..and 10..come before 2 for
>> those who care but ....this is the case when there are you go into a list
>> function and you don't call send.... It just blows up with a 500 ...
>> And it didn't not do this is couchdb 1.2.1...but it does now...
>>
>> originally I thought it was a view problem because it only happened when
>> the keys were of an invalid range... But that was just a symptom.. Not the
>> cause.... So that explains the misleading subject
>> On Apr 17, 2013 12:48 AM, "Benoit Chesneau" <bchesneau@gmail.com> wrote:
>>
>> > So let say it one time for the good. In the current version the
>> > rewriter doesn't parse the variables in nested lists. Maybe it should
>> > be written somewhere in the doc.
>> >
>> >
>> > Some detailed explanation on how work the variable replacement in the
>> > rewriter:
>> >
>> > The current difficulty with variables replacement in the rewriter is
>> > the same yoou have in Erlang. The URL parameters in the URL  are
>> > received as strings. Strings in Erlang are list of integers. Unless
>> > you know it you have no way to know if foo=1 is an integer or a list,
>> > or if true is a boolean or a list. Also if you are passing  "foo" to a
>> > view should it be json encoded or not. The way we bypass that in views
>> > is considering that values given to startkey and endkey are json
>> > encoded. By default all params are url decoded, and json decoded if
>> > needed,
>> >
>> > In the current rewriter we have no expectation on how variables are
>> > passed except for some known parameters (keys, endkeys, startkeys,
>> > end_key, start_key, key) that are considered as json values. To tell
>> > if the parameter "foo" is an integer you have to use the "format"
>> > property.
>> >
>> > Nested list:
>> >
>> > We are for now only replacing the first level in lists [1]. I need to
>> > revisit the code to understand why it does, but I suspect i made a
>> > premature optimization. I will try to fix that asap.
>> >
>> > - benoit
>> >
>> >
>> >
>> > [1]
>> >
>> https://github.com/benoitc/couchdb/blob/master/src/couchdb/couch_httpd_rewrite.erl#L273
>> >
>> >
>> > On Wed, Apr 17, 2013 at 8:42 AM, Stanley Iriele <siriele2x3@gmail.com>
>> > wrote:
>> > > sure...i was using a rewrite handler to rewrite
>> > > /:game/channels/:year/:month/:day/
>> > >
>> > > though these are strings they work for my case and they map to
>> > >
>> > >
>> >
>> db/_design/ddoc/_list/bats/viewname?startkey=[":game",":year",":month",":day"]&endkey=[":game",":year",":month",":day",{}]
>> > >
>> > > this worked just fine before in couchdb 1.2.1 when selecting days that
>> > were
>> > > not populated....
>> > > but the view works fine and returns empty rows...the list function
>> > however
>> > > is where it blew up..aparently if you write start({}) for setting
>> > > headers...but you dont ever call the send function...it blows up with
>> > >
>> > > [Tue, 16 Apr 2013 02:15:20 GMT] [error] [<0.2125.1>] Uncaught error in
>> > HTTP
>> > > request: {error,undef}
>> > > [Tue, 16 Apr 2013 02:15:20 GMT] [info] [<0.2125.1>] Stacktrace:
>> > > [{undefined,write_chunk,[[]]},
>> > >                                  {couch_httpd,last_chunk,1},
>> > >                                  {couch_mrview_show,list_cb,2},
>> > >                                  {couch_mrview,finish_fold,2},
>> > >
>>  {couch_query_servers,with_ddoc_proc,2},
>> > >                                  {couch_httpd,etag_maybe,2},
>> > >                                  {couch_httpd_db,do_db_req,2},
>> > >                                  {couch_httpd,handle_request_int,5}]
>> > >
>> > > but if you just add:
>> > >
>> > > send("") sending an empty string it works just fine..so my work around
>> is
>> > > sending this empty string...it does not pollute the response in
>> > > anyway...but existing code does break
>> > >
>> > >
>> > > On Tue, Apr 16, 2013 at 9:44 PM, Dave Cottlehuber <dch@jsonified.com>
>> > wrote:
>> > >
>> > >> On 17 April 2013 00:29, Stanley Iriele <siriele@breaktimestudios.com>
>> > >> wrote:
>> > >> > nevermind..I found the cause...It is clearly a couhdb bug...but I
>> > found a
>> > >> > work around for it that seems to make everyone happy..if anyone gets
>> > this
>> > >> > error...or even sees this message I'll happily follow up
>> > >>
>> > >> Hi Stanley,
>> > >>
>> > >> Glad to hear you found a solution! Can you share the HTTP query syntax
>> > >> you are using, and also what your workaround is?
>> > >>
>> > >> If it's a bug a JIRA ticket would be welcomed
>> > >> https://issues.apache.org/jira/browse/COUCHDB with the details.
>> > >>
>> > >> A+
>> > >> Dave
>> > >>
>> > >> > On Mon, Apr 15, 2013 at 8:00 PM, Stanley Iriele <
>> > >> > siriele@breaktimestudios.com> wrote:
>> > >> >
>> > >> >> Hi,,,I have just upgraded to using couchdb 1.3.0 from 1.2.1 and now
>> > this
>> > >> >> stack trace is showing up when i query startkey and endkey values
>> > that
>> > >> are
>> > >> >> out of range of the view..is there a flag i need to set or config i
>> > >> need to
>> > >> >> change because this was not hapening before
>> > >> >>
>> > >> >>
>> > >> >> [Tue, 16 Apr 2013 02:15:20 GMT] [error] [<0.2125.1>] Uncaught error
>> > in
>> > >> >> HTTP request: {error,undef}
>> > >> >> [Tue, 16 Apr 2013 02:15:20 GMT] [info] [<0.2125.1>] Stacktrace:
>> > >> >> [{undefined,write_chunk,[[]]},
>> > >> >>                                  {couch_httpd,last_chunk,1},
>> > >> >>                                  {couch_mrview_show,list_cb,2},
>> > >> >>                                  {couch_mrview,finish_fold,2},
>> > >> >>
>> >  {couch_query_servers,with_ddoc_proc,2},
>> > >> >>                                  {couch_httpd,etag_maybe,2},
>> > >> >>                                  {couch_httpd_db,do_db_req,2},
>> > >> >>
>>  {couch_httpd,handle_request_int,5}]
>> > >> >>
>> > >>
>> >
>>