pip show fails silently if package not found

[GadgetSteve]

Environment
All

• pip version:
  Up to 19.2.1

• Python version:
  All

• OS:
  All

Description
When you use pip show somepackage if somepackage is not found there is no output, this is not too bad when a single package is specified but when multiple ones are it is easy to fail to notice that one or more is missing. This was raised as a part of #5454 where trailing commas result in packages not being found.

Expected behavior

I would expect pip show to give a clear indication of any package_(s)_ that it does not find, preferably either at the begining or end of output where it is less likely to be missed as mentioned by @chrahunt in his comments on #5454 - some thing like:

The following package(s) are not installed: somepackage

How to Reproduce

type: pip show asdf asdfasdf gsdfgfg dfgd or any other random set of gibberish then try with 3 packages that you do have installed and one that you don't.

1. type: pip show asdf asdfasdf gsdfgfg dfgd or any other random set of gibberish
2. Then run pip show with some packages that you have and one that you don't or gibberish

Output

> python -mpip show asdf asdfasdf gsdfgfg dfgd 

> python -mpip show asdf requests requests-html
Name: requests
Version: 2.21.0
Summary: Python HTTP for Humans.
Home-page: http://python-requests.org
Author: Kenneth Reitz
Author-email: me@kennethreitz.org
License: Apache 2.0
Location: c:\python36_64\lib\site-packages
Requires: idna, chardet, certifi, urllib3
Required-by: yampy2, wikipedia, Sphinx, safety, requests-html, pypistats, plotly, pandas-datareader, osmnx, nbdime, moviepy, jupyterhub, jupyter-request
s, folium, cookiecutter, caniusepython3
---
Name: requests-html
Version: 0.10.0
Summary: HTML Parsing for Humans.
Home-page: https://github.com/kennethreitz/requests-html
Author: Kenneth Reitz
Author-email: me@kennethreitz.org
License: MIT
Location: c:\python36_64\lib\site-packagesRequires: pyppeteer, bs4, w3lib, requests, pyquery, fake-useragent, parseRequired-by:

Trying to spot the missing asdf in the above is not simple.

I agree with the sentiment that "Errors should never pass silently." and think that this is a case in point.

In the spirit of "Unless explicitly silenced." -q could suppress this message.

[chrahunt]

This seems reasonable to me. I think if we write to stderr then we don't necessarily need to implement a -q, which may be better evaluated in the context of all commands as part of #6099.

[GadgetSteve]

@chrahunt - I agree with leaving out the -q - Looking at #6758 it looks like the output should be to logging.info for consistency with the other code rather than just outputting directly to stderr. _This will currently output to stdout and later switch to stderr as a part of #6758 or some other resolution.