[R] Error Awareness

[Avi Gross]

Stephen,

It is becoming a pattern here.

You have been told R allows ways to check for errors as the code is
interpreted and this DOES NOT distinguish between development aspects and in
the field. It does not matter if a problem is external like a required file
not being in place or having a new name or whatever.

Anything that may go wrong can be handled in quite a few ways. Some are
fairly obvious, such as checking if an expected  list or vector or whatever
is returned that is empty or not set or is the wrong data type or contains
some NA or negative numbers and on and on. All kinds of IF statements and
checks are a way to deal with these and you decide whether to make some kind
of fix or abort.

But obviously if the error happens outside your code (and sometimes also
within) you can use various ways R provides to intercept the error and do
something. Heck, people often use this in advance as a programming tactic
such as using try() or tryCatch() to something like try to open file1 in the
current directory and if that fails, in a specified directory and maybe if
that fails, get it from file2 or maybe standard input ...

Well-written code that wants to be bullet-proof will often be a bit bloated
as it regularly stops and checks not only for all kinds of error conditions
but also checks if the variables being used are in the proper format and
range before being passed to functions that might break, is say you told it
to use a correlation above 1. It also decides to intelligently deal with the
errors, when possible.

I have written code that called a statistical analysis (lavaan)  function
hundreds of thousands of times on randomly generated data and it worked
fine, except occasionally it broke. As I tended to run it overnight, it was
a pain to wake up and find it had crashed. No, I am not aware why it
crashed, but it could have been something as simple as a matrix that could
not be inverted. But for my purposes, it did not matter as what I would be
happy to get is knowing it failed, marking associated rows of a data.frame
accordingly, and filtering out the results in later analyses. So, I wrapped
not only this call but some others as described above and intercepted the
rare problem and dealt with it and let my program continue overnight.

Clearly if you expect anomalies to be rare, this is extra overhead. But if
things do happen, then your code is not robust enough.

-----Original Message-----
From: R-help <r-help-bounces using r-project.org> On Behalf Of Stephen H. Dawson,
DSL via R-help
Sent: Thursday, December 23, 2021 11:17 AM
To: Bert Gunter <bgunter.4567 using gmail.com>
Cc: R-help <r-help using r-project.org>
Subject: Re: [R] Error Awareness

Hi Bert,


Thanks for the reply.

The use case I presented involves working code, not diagnosing a script in
development.

The concern is running a script where something changes outside of the
script, as I stated. An example of a change is a data source changing
perhaps the columns in the data source, or even changing the spelling of a
column. The point is something changes.

The question then becomes how to be aware of the change when the reset of
the script runs fine, as I presented.

Think of having a script running daily where you are not running it
manually. How do you know there are errors in the script processing at this
point?

Does this question now make more sense?


*Stephen Dawson, DSL*
/Executive Strategy Consultant/
Business & Technology
+1 (865) 804-3454
http://www.shdawson.com <http://www.shdawson.com>


On 12/23/21 10:49 AM, Bert Gunter wrote:
> ?tryCatch
>
> This list is not meant for tutorials. A web search on "find errors in 
> R code" brought up what appeared to be many resources to address your 
> question. Please make use of such resources before posting here. Also, 
> please read and follow the posting gui)de linked below to find out 
> what sort of help you can expect (and have already received here.
>
> Cheers,
> Bert Gunter
>
> "The trouble with having an open mind is that people keep coming along 
> and sticking things into it."
> -- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )
>
> On Thu, Dec 23, 2021 at 3:14 AM Stephen H. Dawson, DSL via R-help 
> <r-help using r-project.org> wrote:
>> Hi,
>>
>>
>> I am thinking about awareness of errors when an R script runs.
>>
>> My concern is I have an error-free script. I run it for months on end 
>> without problems. Then, something changes somewhere causing an error. 
>> My wonderment is how R will tell me I had an error in the script, but 
>> the rest of the script ran without impairment.
>>
>> QUESTIONS
>> What are some of the more helpful options available to an R developer 
>> to capture errors in a script run?
>>
>> What are some of the best processes to implement these more helpful
options?
>>
>>
>> Thanks,
>> --
>> *Stephen Dawson, DSL*
>> /Executive Strategy Consultant/
>> Business & Technology
>> +1 (865) 804-3454
>> http://www.shdawson.com <http://www.shdawson.com>
>>
>> ______________________________________________
>> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see 
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide 
>> http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.

______________________________________________
R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.