Constructing gtools and ftools ado packages remotely with no internet, and continuously encountering error r(111)

[Jakob Brounstein]

Hi all,

I'm trying to remotely construct gtools and ftools on a computer that has no internet access with the help of an on-site research assistant. A bit of a logistical nightmare.

The Microsoft computer has run gtools/ftools commands before, but Stata was recently upgraded to Stata 18 from 17, and the ado directories were replaced.

My problem is that on many of the gtools subcommands, like "fasterxtile" and "gegen" and "gduplicates", I keep on getting the error r(111).

This is the error that pops up for "variable does not exist" or "ambiguous abbreviation". I know for a fact that the variables I am referring to DO exist and are NOT ambiguously abbreviated (I have set varabbrev off).

I have tried installing ftools and gtools locally and transferring over the ado/mlib/plugin files to the local ado folder, and I have run mata: mata mlib index

But I keep on getting this error r(111).

For instance, I have id variables (I have an id-time panel), and just to check the syntax, I have run (for example)
gduplicates drop id, force bys id: gegen check = mean(id) Here you can find images of me running "gduplicates drop" with and without trace on.

And both of these commands output r(111).

Does anyone have any advice? Are there likely ado/mlib packages that I am missing?

I am VERY desperate at this point.

[George Ford]

shot in the dark here.

https://github.com/mcaceresb/stata-gtools

on the "code" button, you can download a zipfile.

unzip it to a directory and then try this for a local drive install.

https://kb.iu.edu/d/azwf

[George Ford]

A possible solution here.

https://www.statalist.org/forums/forum/general-stata-discussion/general/1353234-installing-packages-without-internet-access

[Jakob Brounstein]

Thank you so much! Along this vein, I think I'm going to try to just download the entire gtools + ftools build folders and do a net install. I guess the answer seems obvious, as I was previously just copying ado files from my plus folder.

I'll report back whether it worked.

[Nick Cox]

Cross-posted at https://stackoverflow.com/questions/...rnet-encounter and reddit.com/r/stata/comments/15eq45i/…

Telling people about cross-posting is (1) a rule on Reddit (2) a request here (3) a good idea on Stack Overflow.

I can't add to advice already given.

[Jakob Brounstein]

My bad Nick Cox ---I now clearly see those rules now. Thanks for pointing that out.

To close the thread, I've solved this problem just by touching as little of the packages as possible and doing a local net install.

1. I went to the respective github sites for both ftools and gtools and downloaded as folders "https://github.com/mcaceresb/stata-gtools/tree/master/build" and "https://github.com/sergiocorreia/ftools/tree/master/src", naming them "gtools_build" and "ftools_src"

1a. I had to figure out which folders on the github sites could be used to compile gtools and ftools via net install in Stata. I just did some local troubleshooting to determine that these were the necessary folders (as opposed to the other ones like "https://github.com/mcaceresb/stata-gtools/tree/master/src")

2. Transferred the zipped folders to my on-site computer using an email service other than gmail (as gmail doesn't allow one to send .plugin files)

3. Unzipped the zipped folders on my on-site computer (remotely with my research assistant)

4. Performed a local net install as

net install ftools, from("[MYPATH]/ftools_src") replace
ftools, compile
net install gtools, from("[MYPATH]/gtools_build") replace

And now they work fine. Moral of the story I guess is to try to as little manually as possible (I was previously manually copying over the .ado, .mata, etc. files from gtools and ftools)