Benno
@benno@eigenmagic.net
You're absolutely right!
0
Followers
0
Following
Joined August 19, 2018
Website:
GitHub:
Posts
Open post
Open post
Idle thought: In the 1990s I wrote a lot of Perl. Then I switched to Python and found I enjoyed writing that a lot more, and this persisted in that I also tried Ruby and didn't like it.
One of the things I'd come to realise is that while the idea of a language like Perl or Ruby that tried to be "expressive" or, to a certain extent, closer to natural language than other programming languages were/are sounds nice and narrows the gap between thinking something and expressing it as code, it allows the ambiguities of natural language to creep in to code where ambiguity is Really Bad.
Relatedly I've never liked natural language interfaces to things. I find it adds an awkward translation layer between the goal I have in my head and the achieving of it. Explaining things to other humans directly is fine, since I understand the need to communicate for a group to achieve an end, but I know the computer is just going to turn it into something else and so, again, the ambiguities show up, and so I'm now worrying about the way to phrase what I want to something where ideally I can implement my goal much more directly without having to get English involved.
So yeah, part of what puts me off using LLM tools is I do _not_ want to talk to them. I don't want to spend the effort turning my thoughts into prose for them. I'm not _in_ natural language mode at that point, and I don't want to be. So are there ways to use these things that don't require that mode of interaction?
One of the things I'd come to realise is that while the idea of a language like Perl or Ruby that tried to be "expressive" or, to a certain extent, closer to natural language than other programming languages were/are sounds nice and narrows the gap between thinking something and expressing it as code, it allows the ambiguities of natural language to creep in to code where ambiguity is Really Bad.
Relatedly I've never liked natural language interfaces to things. I find it adds an awkward translation layer between the goal I have in my head and the achieving of it. Explaining things to other humans directly is fine, since I understand the need to communicate for a group to achieve an end, but I know the computer is just going to turn it into something else and so, again, the ambiguities show up, and so I'm now worrying about the way to phrase what I want to something where ideally I can implement my goal much more directly without having to get English involved.
So yeah, part of what puts me off using LLM tools is I do _not_ want to talk to them. I don't want to spend the effort turning my thoughts into prose for them. I'm not _in_ natural language mode at that point, and I don't want to be. So are there ways to use these things that don't require that mode of interaction?
34
4
13
0
Open post
I just ran across a situation where some code-generating entity, human or otherwise, created a Python file with dashes in the name, and then in another file, decided they needed to import it and do a bunch of gymnastics to get around the dashes rather than, y'know, renaming the original file...
10
1
1
0
Open post
In reply to
@cuddlyanarchist You must fight me for at least five minutes before I will agree.
2
0
0
0
Open post
Make me Smart Glasses but the only thing they do is replace all advertising around me with black boxes containing the word "ADVERTISEMENT" in small letters.
25
2
11
0
Open post
RE: https://mastodon.social/@nowaynews/116363735967942248
Maybe @mwl will know what to do with this information?
Maybe @mwl will know what to do with this information?
2
0
1
0
Open post
Open post
I'm the guy who decides whether Markdown links are [title](url) or (title)[url] and I change my mind every few days. All implementations of Markdown immediately follow my lead.
Ask me anything (I will ignore it).
Ask me anything (I will ignore it).
31
2
11
0
Open post
Open post
In reply to
@liamvhogan Did you do something stupid like kill its girlfriend on Devil's Night?
1
0
0
0
Open post
In reply to
@abstractcode I don't see a problem with a talk like that at all, or if I did it'd be making sure you weren't getting yourself into legal trouble. I mean hell PyCon AU has had the "DevOops" specialist track several times which is the natural home for that kind of talk. The "attacking yourself" thing only hits for me when you're using it as a way to say "it's ok for me to make fun of them because I also made fun of myself" which is never how it works.
Agree 100% on the code of conduct point. There's no harm in being explicit in the behaviour you expect.
Agree 100% on the code of conduct point. There's no harm in being explicit in the behaviour you expect.
2
2
0
0
Open post
In reply to
@abstractcode And to be 100% clear: I'm not directly involved in the organising of PyCon AU this year. It's possible I could end up helping with proposal review (as a normal member of the review team) but I'm not on the committee. This was just me writing a "guest post" to help out.
1
2
0
0
Open post
In reply to
@abstractcode No specific incident, although I don't mind saying that I've seen a few "Make [thing] Great Again" titles in various CfPs, mine and others, and it's always been something where my position has been "if this gets in it will be with a different title".
It's more in the spirit of being explicit about what our position on these things are so that people know what they're dealing with and what the expectations are. So many issues with humour come from a place of lack of thought rather than anything actively malicious and so this helps people know what they should be careful with and what questions to ask themselves.
The ones that are being actively malicious usually out themselves in other ways, tbh.
It's more in the spirit of being explicit about what our position on these things are so that people know what they're dealing with and what the expectations are. So many issues with humour come from a place of lack of thought rather than anything actively malicious and so this helps people know what they should be careful with and what questions to ask themselves.
The ones that are being actively malicious usually out themselves in other ways, tbh.
2
6
0
0
Open post
I wrote a thing for @pyconau on being funny.
https://mastodon.pycon.org.au/@pyconau/116167724589175889
https://mastodon.pycon.org.au/@pyconau/116167724589175889
14
2
16
0
Open post
Enormous ute? That's a paddlin'.
Nothing in the tray? That's a paddlin'.
It's a Ranger? That's a paddlin'.
It's matte black? You'd best believe that's a paddlin'.
Nothing in the tray? That's a paddlin'.
It's a Ranger? That's a paddlin'.
It's matte black? You'd best believe that's a paddlin'.
33
0
12
0
Open post
In September of this year it will be the 26th anniversary of Lobster Magnet: https://archive.org/details/lobster-magnet
1
0
1
0
Open post
So I was invited by the lovely folks at @pyconau to give a keynote. I think it went pretty well. I’m sure it won’t annoy anyone this time. http://youtu.be/yc248GXjvA8 #PyConAU
33
2
25
0