Re: st: Unreasonable error "Obs. nos. out of range"

[Nick Cox <njcoxstata@gmail.com>]

You have not answered all my questions about your code.

I don't know what your programming criteria are here.

I note only as one common example that references to [_n-1] and [_n+1]
are commonplace and deliberate.
I don't want to see error messages that myvar[0] and myvar[_N+1] don't
exist. I know that already.
Nick
njcoxstata@gmail.com


On 17 June 2013 17:36, Stefan Bernhard <stefanbernhard88@gmail.com> wrote:
> Thx Nick, so the thing ist that it is actually possible to refer to a
> non-existant observation, which caused the confusion.
>
> I then found the error after I knew that the observations had to be
> diasspearing.
>
> But I think the way this is handled is somewhat awkward. It should
> already tell me that there is no 2nd observation when I try to refer
> to it with the if clause :/
>
> best regards, stefan bernhard,
>
>
>
> 2013/6/17 Nick Cox <njcoxstata@gmail.com>:
>> Difficult to comment given this little information, but
>>
>> 0. It is possible to have variables defined but no observations.
>>
>> 1. Possibly more to the point, the -replace- will certainly fail if _N
>> < 2 but it's not necessarily an error otherwise to refer to -cats[2]-
>> when it doesn't exist. If -cats- exists as a variable then any
>> references to subscripts that don't correspond to observation numbers
>> are interpreted as missing.
>>
>> 2. However, your code does imply that -cats[2]- is being treated as
>> zero. Could you confirm that -local bak- is not defined before the
>> code you cite?
>>
>> Either way, -list-ing the data would make your situation clearer.
>>
>> This example bears on #1.
>>
>> clear
>> set obs 1
>> gen cats = 1
>> if cats[2] == 0 {
>>       di "problem 1"
>> }
>> else di "problem 2"
>>
>> Nick
>> njcoxstata@gmail.com
>>
>>
>> On 17 June 2013 15:21, Stefan Bernhard <stefanbernhard88@gmail.com> wrote:
>>> dear statalisters,
>>>
>>> i have a piece of looping code over different variables and all
>>> observations, and an excerpt of the trace shows this:
>>>
>>> = if cats[2] == 0 {
>>>   local bak = 0
>>>   }
>>> - noi di as text "bak is `bak'"
>>> = noi di as text "bak is 0"
>>> bak is 0
>>> - replace `var' = 1 in `i'
>>> = replace cats = 1 in 2
>>> Obs. nos. out of range
>>>
>>>
>>> This makes no sense at all to me.
>>>
>>> In the first line, it successfully uses the value of cats of
>>> observations number 2 to define the local bak as 0.
>>>
>>> Few lines later, it acts as if there was no more observations number 2
>>> and cannot replace the number of cats with 1 in observation number 2.
>>>
>>> Why does it say Obs. nos. out of range ?
>>>
>>> regards, stefan bernhard,
>>> *
>>> *   For searches and help try:
>>> *   http://www.stata.com/help.cgi?search
>>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>>> *   http://www.ats.ucla.edu/stat/stata/
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>> *   http://www.ats.ucla.edu/stat/stata/
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/