What does _rc 111 mean in merge?

[Daniel Feenberg]

I have a merge statement that returns 111 return code, but appears to have worked.

merge 1:m _taxsimid state year using `original'

Result                           # of obs.
    -----------------------------------------
    not matched                             0
    matched                                 1  (_merge==3)
    -----------------------------------------
Merge failed 111

                 _merge |      Freq.     Percent        Cum.
------------------------+-----------------------------------
            matched (3) |          1      100.00      100.00
------------------------+-----------------------------------
                  Total |          1      100.00

Note that all (one) of the records matched, and I can list the record to confirm that the merge was done correctly. So what could be the problem causing the "Merge failed 111" error message?

[Leonardo Guizzetti]

Hmm...in the manual section [U] Chapter 8 Error Messages and Return Codes, section 8.1, error code 111 is listed as an error when no variables are defined (or found by that name).

Also, error codes 100-199 are reserved for syntax errors (see the help file behind -help error-). The following possibilities are listed:

• variable(s) not found/defeind
• variable(s) out of order
• variable(s) not found in using data
• ambiguous abbreviation

Could it be that one of your variables among -_taxsimid state year- were not found in the using dataset? Perhaps you could post a reproducible example?

[Daniel Feenberg]

Answering my own post - 111 seems to be the default value for _rc until -capture- is executed. I forgot the -capture- on the -merge- statement, as is evident in the code snippet posted.

. di _rc
111

. capture di " "

. di _rc
0

[daniel klein]

111 seems to be the default value for _rc until -capture- is executed

No, it is not. Every session of Stata starts with the return code set to 0.

. display _rc
0

. capture display " "

. display _rc
0

I second Leonardo's query: please provide a reproducible example(s).

Best
Daniel

[Leonardo Guizzetti]

Answering my own post - 111 seems to be the default value for _rc until -capture- is executed. I forgot the -capture- on the -merge- statement, as is evident in the code snippet posted.

I second daniel's response to this. Per the help file for -capture- (emphasis mine).

capture executes command, suppressing all its output (including error messages, if any) and issues a return code of zero. The actual return code generated by command is stored in the built-in scalar _rc.

I'm almost positive there's problem with your data setup.

[Bill Rising (StataCorp)]

Could you post enough data code to reproduce the "Merge failed 111" output? It is very odd.

As for the _rc code:

111 seems to be the default value for _rc until -capture- is executed.]

This isn't quite right. If you start Stata from scratch and try

display _rc

you'll see that _rc is 0 to start with.

_rc retains its value from one -capture- to the next regardless of the return codes of any commands in between, so _rc == 111 means that the most-recent -capture- caught a return code of 111:

clear
* capture an r(111), since hmmmm does not exist:
capture summarize hmmmm
display _rc
* a command that works, and hence has a return code of 0:
display "This is just fine"
* _rc still has the value from the last -capture-:
display _rc

[Daniel Feenberg]

I suppose it isn't reproducible on your system, it must be something special about our system, but here is my test case cut-and-pasted from my screen. The sysprofile.do executes without error when executed by itself but there is no profile.do, so attempts to execute that return r(601). Neither does a -capture- so neither should affect the _rc if the documentation cited above is applicable.

Notes:
      1.  Unicode is supported; see help unicode_advice.
      2.  More than 2 billion observations are allowed; see help obs_advice.
      3.  Maximum number of variables is set to 5000; see help set_maxvar.

running /disk/admin/Linux-7-local/stata15/sysprofile.do ...

running /disk/homedirs/nber/feenberg/profile.do ...

. di _rc
111

. capture di _rc

. di _rc
0

This was run on Statamp 15.1 on a Linux 64-bit system. It is consistent.

Daniel Feenberg

[daniel klein]

Well, whatever is causing the return to be set to 111 must be inside

/disk/admin/Linux-7-local/stata15/sysprofile.do

and/or

/disk/homedirs/nber/feenberg/profile.do

We cannot see either content, so we cannot comment on that. My guess is that one of those files contains a capture statement that catches return code 111.

However, I do not see how any of these files could possibly produce the error message

Merge failed 111

that appears to be specific to merge (although the capitalization seems odd, too).

Best
Daniel

[Daniel Feenberg]

My sysprofile.do contains several -adopath.ado- commands. This is an unmodified -adopath.ado- as supplied with Stata. None print an error in the log when executed, but there are capture commands in -adopath.ado- that do not print a message for non-zero returns, perhaps because the situation is not an error. The matter is closed.