Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 16 Sep 2016 15:11:51 -0400
From:      Jim Ohlstein <jim@ohlste.in>
To:        Warren Block <wblock@wonkity.com>, ports@FreeBSD.org
Subject:   Re: Checking port option descriptions
Message-ID:  <3de26d31-e4e0-ddc4-27ae-03bab473849b@ohlste.in>
In-Reply-To: <alpine.BSF.2.20.1609160951090.12548@wonkity.com>
References:  <alpine.BSF.2.20.1609160951090.12548@wonkity.com>

next in thread | previous in thread | raw e-mail | index | archive | help
Hello,

On 09/16/2016 11:52 AM, Warren Block wrote:
> Ports options ask the user to make a decision on whether to enable that
> option.  Option descriptions are critical for this, giving the user
> information to help them make that decision.
>
> Unfortunately, what is clear to the porter is often not clear to a user.
> The Porter's Handbook says "Do not just repeat the name", but this still
> happens, either exactly, or with a description that adds no information.
>
> For example:
>
>   XYZ    Enable XYZ
>
> The description here adds no information. The name of the option itself
> tells the reader that this is for enabling or disabling a feature. The
> option asks them to make a decision, whether to enable that option or
> not, or even just to leave it at the default, but does not give them any
> help in making that decision. Let's improve that:
>
>   XYZ    Include protocols for use with XYZ servers
>
> This gives the reader some additional details.

"[S]ome" being the operative word here. I don't disagres with your basic 
premise, but the truth is, at the end of the day it's up to the user to 
understand the consequences of his decisions. If a user doesn't know 
what 'XYZ' is, then adding 'Include protocols for use with XYZ servers' 
probably doesn't tell him or her that much. On the other hand, if a user 
knows what 'XYZ' is, then 'Enable XYZ' is likely enough information with 
which to make a decision.

So in this case there are likely to be two categories of users: those 
who know what 'XYZ' is and those who do not. Those in the former have 
the information either way. Those in the latter have three basic choices:

1) Educate themselves before possibly adding software to their system 
that they do not fully understand, thereby moving into to the former 
category.

2) Choose the default, on the (very possibly mistaken) assumption that 
the porter "knows what's best." Unfortunately that assumption may be a 
bad one, as the porter/maintainer is more likely to choose something 
that satisfies "most users" and loads people with unnecessary 
dependencies (thus defeating much of the benefit of building your own 
ports), or worse, to choose options that work best for him or her.

3) Ask themselves Harry Callahan's famous question, "Do I feel lucky?" 
and go away from the default.

>
>
> Because so many of the option descriptions have predictable
> no-added-information styles, it is possible to write a program that
> detects these. In the process of doing that, I found some actual bugs in
> descriptions that were not caught by other parts of the ports build or
> portlint.
>
> The program is called optcheck and can be found here:
>   http://www.wonkity.com/~wblock/tmp/optcheck/
>
> The readme.txt explains a little more, and optcheck-output.txt is a full
> run against the ports tree from a couple of weeks ago.
>
> The tests are just some that I came up with quickly, and can certainly
> be improved. More can be added, and they can make better suggestions.
> Ultimately, the hope is that this functionality will be added to
> portlint or somewhere else that would help prevent these types of
> pointless descriptions.
>
> For usage, run
>   optcheck -h
>
> For a run against the full ports tree, use just
>   optcheck
>
> To run against a category or single port directory, use -d:
>   optcheck -d /usr/ports/devel
>   optcheck -d /usr/ports/print/ghostscript9-base
>

-- 
Jim Ohlstein



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3de26d31-e4e0-ddc4-27ae-03bab473849b>