Front page | perl.perl5.porters |
Postings from July 2014
RE: New feature proposal : <<>> to disable magic open of ARGV
Thread Previous
|
Thread Next
From:
bulk 88
Date:
July 27, 2014 03:42
Subject:
RE: New feature proposal : <<>> to disable magic open of ARGV
Message ID:
COL126-W496A78D4185CEA7A074A6DDFFA0@phx.gbl
> Date: Sat, 26 Jul 2014 18:38:03 -0400
> From: perl.p5p@rjbs.manxome.org
> To: demerphq@gmail.com
> CC: rgs@consttype.org; perl5-porters@perl.org
> Subject: Re: New feature proposal : <<>> to disable magic open of ARGV
>
> Language changes are things that have to be designed, and that means
> discussion.
>
> Here's my counter-proposal on bike-shedding: no more using the phrase
> bike-shedding. It's now used to mean "somebody wants to talk about something
> where I think I'm already right," and it's too emotionally charged. If you
> think someone is debating pointless trivia, suggesting changes of no
> consequence, or can't see the forest for the trees, point out the specific
> problematic behavior.
Too many language design threads, the so called "bike shedding" are a
waste of my time to read. I stop reading after 10 screenfuls of text. Many
of these threads turns into mental battles of flawed analogies, or "X
does it as Y, we need to do Y", or "my day job [and therefore all darkpan
code] does Y", or just plain I'm going win the game by arguing, but not
enough arguing to get me locked up and arraigned for violating SOC.
A number of people (not naming names) who go into language design
threads, I never see their names except in language design.
They do not otherwise participate in P5P, no tickets, no ticket comments,
no patches. There isn't really a way to legislate a way to identify
non-productive posters, since their posts are not filled with ad hominums
or pure fiction. Calling people out for non-productive posts will likely
be interpreted as personal attack on its own. If well known committers or
rjbs get involved, now its the p5p cabal scheming to silence the
"non productive" poster.
Person A is a non-productive poster by most people. Person B is a complainer
/accuser/moderator. Person B makes a most in a thread where person A posts alot
saying person A is making non-productive junk posts. Person A then writes
back saying person B is being non-productive, Person B is "rude" and needs to learn
manners. Person B isn't a team player. Person B is trying to break up the discussion
which is going the way person A likes, and person B needs to be removed from the
thread and should have never entered it in the first place. I've seen a couple threads
in the Perl world, where someone responded to every last fucking post in the thread.
No exceptions. It is either a you are welcome/thanks for supporting my idea post, or
a wheelbarrow dump of rehased criticism to make them see the light
( see http://xkcd.com/386/ ) (I think it is a sign of forgetting to take their meds lol).
Arguments from 1-2 weeks earlier start
getting repeated at 6-15 messages deep. Some people post the same
counterpoint/argument in reply to 2-3 messages, so 3 messages by the
same person have the same argument in them, it was a waste of my time to click
them. Personally, when I see I am writing the same response to multiple thread
posts, I will copy paste the quoted text from all the relavent messages, and just
make 1 email, and 1 paragraph addressed to the authors of 3 different
messages. The language design threads look like a chatroom,
but with each line taking a dozen hours instead of 10 seconds. The
threads move so slowly, I have no idea after 4 days or 1 week what it
is about anymore, and I am not going to reread 25 messages (which
I already read and forgot a week ago) to find what is going on. One
battle strategy I see is, if you say it enough times, it becomes true.
That is a waste of my time to read. The davem grant reports have hours
of p5p reading on them. If that is what he does, how many non-paid
"volunteers" are spending an equal number of hours reading p5p
ML instead of debugging/writing code?
The only way to make sense of these language design threads is a
wiki with pros and cons and a moderator that will de-duplicate arguments
into at most 2 (preferably 1) screenfuls of information. Optionally recording
some kind of collapsible list (+/-) of nicks that vote for or against (maybe
the UI is "for votes" only) the argument. Also it isn't clear without running
code, what the proposed feature does currently? Syntax error? runtime error?
which one exactly? I want bullet points, not deeply nested threads. What is
the interp overhead for this feature for people that will never use it?
To derail one of these threads, start adding features. Also most of the posters
and commentators are PP posters, not perlguts posters. When I see a
language design request, even though IDK enough to instantly come up
with a "cost" for the feature, it worries me how much more C logic needs
to be added to implement it. And of course the C code won't say what PP feature
this C block/branch is implementing, and it will be added without ever looking
at the code above or below the insertation point. Identifiers like "my", "really" and
"special" like to sneak in, which explain nothing, as the function implements more
and more features. Then functions grow to the size of a small virtual machine
in LOC. Try understanding pack() or open() for example. Without patches, and
for a feature in a part of PP I never use it is difficult for me to evaluate, or want to
spending hours reading to come up with an opinion and usually a 1-3 sentence
answer. 3 hours to write 3 sentences is inefficient.
One part of warnocking is, nobody cares, so they stay silent. Should I (bulk88)
really write a response that says "I dont care"? Are such responses useful in forming
consensus? Should silence/IDC be a no vote by policy? Is silence/abstain/IDC a
politically correct version of the "no" answer that should be treated as what it is,
a "no" vote?
This post took me 57 minutes to write, but I hope it will reduce the minutes I need to
spend on future design controversies.
Thread Previous
|
Thread Next