[R] (Off-Topic) Time for a companion mailing list for R packages?
@v|gro@@ @end|ng |rom ver|zon@net
Fri Jan 14 01:07:14 CET 2022
I appreciate your thoughts and some history, Marc.
My personal goal here is to both learn and if possible, be helpful.
I am just bothered when one or two people regularly chime in to remind people that this forum should be pretty much only about base R. One way to satisfy them is to move some discussion away but another is to advise them the (un)official rules are changing a bit and will tolerate just a bit more.
I think the R STUDIO distinction is next to meaningless. Almost every package that is commonly used can be used anywhere in the R universe. You do not need to be in R Studio to use ggplot or dplyr or glue or many other things that happen to be loosely packaged in the tidyverse. If someone asks a reasonable question about why their multi-line ggplot code did not produce quite the output they expected and someone replies that ggplot uses the current ordering of factors so simply change that before calling it or sort of inline by calling a function that makes a re-ordered copy, so what? In a sense it turns out to be a question about base R and how you create and change factors.
I think though that a good attitude for people answering here should be to begin by mentioning if something the user asks for can easily enough be done in base R and perhaps supply a hint as to what can be used. They might (or might not) then point out that some other packages like data.table have some functionality built-in or are reputed to be much faster, or that doing it using dplyr tools and a pipeline is arguably easier to understand and uses more powerful features if needed, ...
Years ago I learned plenty about base graphics and lattice and then ggplot. Do many people ask about lattice and is it used that much? It does not seem to come up and I suspect that is not because base graphics meets all needs, but because ggplot overall provides enough overlapping functionality and more, ...
Note I had to temporarily switch mailers and use the AOL mailer in my browser and I have not found any easy way to switch to text-only mode on a per-message basis.
From: Marc Schwartz via R-help <r-help using r-project.org>
To: Eric Berger <ericjberger using gmail.com>
Cc: r-help using r-project.org <r-help using r-project.org>
Sent: Thu, Jan 13, 2022 3:08 pm
Subject: Re: [R] (Off-Topic) Time for a companion mailing list for R packages?
With respect to your second point, as one of the moderators for the
R-Devel list, where the same considerations apply to all "official" R
lists like R-Help here, the issue of plain text content restrictions is
multi-factorial, partly with security considerations in mind, but has
historically been imposed by the university (ETHZ) that hosts and
archives all of these R related lists. It is also, historically, very
common with open source communities generally.
If one sends a message to the list that is only HTML, or perhaps a
multi-part MIME message with both HTML and plain text sub-parts, the
HTML markup will be stripped, leaving only the plain text part. You will
usually see a line at the end of the post/reply as follows:
"[[alternative HTML version deleted]]"
when that occurs.
When we see some messages here that appear to be a single, longish,
unformatted and largely unreadable paragraph, with no obvious line/para
breaks, where the sender may have actually formatted the message in HTML
before sending, that is the result of the HTML filtering by the ETHZ
server, perhaps also influenced by how the sender's e-mail client
formatted the plain text sub-part of the message.
That is why it is best, and recommended, to format the e-mail message
natively in plain text only to begin with, and there are numerous e-mail
clients that support that functional distinction.
Also, there are still folks out there that use plain text only e-mail
clients, yes in 2022, where we have to recognize and be sensitive to
that reality as the official R support venues.
With respect to attachments, a limited set of them are allowed here for
the R lists. Primarily, these are PDF, [E]PS, PNG and JPG, which would
generally accommodate the graphics related scenarios that you mention.
As has been noted elsewhere in this thread, there are a number of other
venues that provide R specific, community based, support using online
forums with varying levels of, perhaps more "modern" functionality. The
most prolific likely to be StackOverflow which, without any doubt, and
to some extent via the gamification on that platform, has markedly
reduced the traffic volume here over the years. That shift initially
started in circa 2009 when that platform began, and then rapidly grew,
with traffic volume on R-Help peaking in 2010. I occasionally run
traffic volume reports comparing the two venues over time, and having
been a part of the R community for ~20 years now, the drop-off here post
2010 is quite noticeable. I don't label that phenomenon as good or bad,
it is simply the reality of the situation and this is not a competition
As noted in this thread, other locations find R related interactions as
well, which would include LinkedIn, ResearchGate, RStudio's forums, and
some package developers provide support via their GitHub repos, using
the Issues functionality, even if not a bug. In the case of RStudio,
having their own support forums, and as comes up here, there is
frequently a lack of distinction between RStudio the third party GUI,
and R itself, resulting in confusion by users, and subsequent
"energetic" discussion here.
I am not advocating the general use or non-use of any specific platform,
but am primarily addressing the limitations here on the official lists
as currently implemented. For the record, I am not an RStudio user and
do not participate on StackOverflow, albeit, in the latter case, Google
searches on technical subject matter do lead me there on occasion.
We have had numerous online and offline discussions over the many years
of sub-setting the official R lists by common subject matter. The
challenge is always, how much granularity do you impose on the list
structure, even if you just create an "R-Newbie" list, which
specifically has been discussed many times over the years. The more
granularity you impose, the more difficult it becomes to manage and
interact with those "stove pipes", when there may not be a clear
separation in subject matter, and there will be the inevitable overlap
in content. You also need relevant, volunteer, community members to then
be willing to moderate, subscribe to and participate on those multiple
lists, if they are to be of value.
That would be the challenge even with an R-Packages list. Will every, or
most, or the most popular, third party package developers subscribe to
that list, so that if someone posts a query, they are likely to get a
timely reply? Would enough community members with relevant experience do
so? Or would a useR then re-post the query to R-Help after a period of
time of no replies anyway, and in time, give up on the other list due to
the lack of traffic?
For many third party packages, especially those that have a longer
history and have a reasonable useR base, posting here is fine in my
mind, and helpful community replies are reasonable to expect. For
others, communicating with the package maintainer directly may be the
only recourse, because there is perhaps not broad experience with the
package in the community. It may be difficult for the useR of such a
package to make that distinction a priori, unless they (hopefully) take
the time to search the list archives first, perhaps using rseek.org or
Google, which is also recommended.
I always begin with the assumption that the question that I have has
been asked before, and hopefully an answer will also be present. That
may be in the relevant documentation, including package vignettes where
available, or otherwise online here or elsewhere. Certainly, if there is
a bug or a reasonable suspicion of one in a third party package, the
package maintainer should be the primary point of contact.
Eric Berger wrote on 1/13/22 12:25 PM:
> Re: constructive criticism to make this list more useful to more people:
> Suggestion 1: accommodate questions related to non-base-R packages
> This has been addressed by many already. The current
> de facto situation is that such questions are asked and often answered.
> Perhaps the posting guide should be altered so that such questions fall
> within the guidelines.
> Suggestion 2: expand beyond plain-text mode
> I assume there is a reason for this restriction but
> it seems to create a lot of delay and often havoc. Also, many questions on
> this list relate to graphics which is an important part of R (even base R)
> and such questions may often be more easily communicated with images.
<...earlier content snipped...>
R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
[[alternative HTML version deleted]]
More information about the R-help