> I'd like it if the Air was about half the size. [...]
> I'd like it if the HD display was bigger and much lighter and didn't have such a massive octopus of cables attached to it.
Looks like Apple agreed. The more recent Apple displays are bigger and have mostly eliminated the “octopus of cables” (when LightPeak hits this should even further improve), and the new Air is substantially smaller.
I'm thinking about getting a new Macbook Air, but I'm concerned about his desire for unlimited fast CPU: the new Airs have pointedly lacking CPU resources.
One common thing in all usesthis.com interviews is, no one uses Windows OS as their main operating system and Mac is used most. Only person I found using Windows is Mitch Altman so far.
I do the meat and potatoes of my work in MS Visual C++ Express. I find, the best development environment is the one that you've been using for 10 years. When I do iPhone builds, I first do my best brontosaurus impression, then I copy all my code and art onto a memory stick, unplug my monitor, mouse and keyboard from my PC and then plug everything into my Mac.
The indie games scene is still pretty PC focused because of the long standing domination of the PC over the Mac in the gaming area. It's only now slowly changing.
Yet, 90% of developers use Windows on a daily basis. The people interviewed in usesthis are just not very representative of the community (a bit like using a conference to claim that 90% of developers use Mac Books).
I'm still sad to see some of these people (including pg) use such antiquated tools. I use vi on a daily basis but just for very small text files, you can be so much more productive with more modern tools these days.
In the Unix culture the people who claim to use vi actually use vi + shell + hundreds of little software tools. If it have worked for 30 years perhaps you should give it a try.
>That's not a very good argument. Horse-drawn buggies worked for well over 30 years.
I can't see how this is a good counter-argument either. They still do work, they run on renewable fuel, growing the fuel can be enhanced by the waste products of the horse. They're relatively low maintenance, the motor can be interchanged and easily used to drive other machines. Certainly at city driving speeds the buggy is going to compete quite well.
Yes a car or other automobile is going to outperform a horse & buggy in many situations but there are other aspects that are wins for the horse & buggy.
I'd be interested to see a running cost and efficiency comparison, shoeing horses is probably quite expensive nowadays with the dearth of blacksmiths/farriers.
TextMate is pretty popular and very productive, and it basically goes w/ the vi model. You extend it using external scripts, i.e. a bunch of little helper programs.
This may not be historically accurate - ie not what Bill Joy originally intended, however I think of the "vi model" as the modal editing style which is what differentiates vi/vim for me.
"Use Windows" might to be the key here. I mean, my main development machine is a MacBook Pro, and I use headless Linux machines heavily. But I've always got a Windows machine booted up, and I've got Parallels running Windows on my MBP about half of the time.
I find that 90% little hard to buy. Based on my experience, I would guess they all (Mac/Linux/Windows) have roughly 1/3rd share, the trend being rising for Mac and downward for Windows. (Okay, my view may be Western centric.)
Is 'vi' mature? Or are you talking about Vim? vi has been around a long time, but the functionality is rather limited compared to Vim. When was the last development on vanilla vi (not Vim in compatibility mode)?
I was probably being a little pedantic, but 'vi' can mean:
- The original 'vi' editor
- The entire 'vi collection.' By that I mean vi + clones (i.e. vi, vim, elvis, etc).
- The Vim editor (being the most popular vi clone)
I'm never sure which people mean (though they usually just mean Vim), and I've come across many people that are confused (especially if they aren't too familiar with the Unix/Linux ecosystem and/or are Emacs users) but the inter-changeable usage of terms.
But 90% of developers produce negative work with these tools. It's not very useful to see what tools people use to produce crap. It's interesting to see what tools people choose to optimize their own creative productivity.
I use vi on a daily basis but just for very small text files, you can be so much more productive with more modern tools these days.
I use Windows on a daily basis because I work in a place that insists I do, like most large organisations. 90% usage, if true, doesn't indicate Windows is better or worse.
(Given the choice I would do my day to day job on a Mac.)
yeah, he motivated me to get a thinkpad over a macbook at the time.
That said, i'm getting tired of my T500 already. I don't see myself really needing expresscard as much as I was bent on it.
Overall, I've learned that ergonomics matters at the end of the day no matter what machine you use - meaning I care less about laptop screen size and more about bang for the battery life. I can't convince myself to buy a desktop either, so I'm considering getting a 13-14 inch laptop with the fastest processor, ssd, and gobs of ram instead.
I know this isn't a popular POV, but I've always thought that developers should work on last gen hardware. Getting good performance out of last gen hardware ensures great performance for the typical user.
(okay, let's be reasonable, the build machine should be stupid fast so nobody is sitting around for half a day while the repository builds).
I think that would be true for testing, but not for actual development. When I'm working I want to marginalize anything that might slow me down. You did mention the build machine should be fast - but don't most people build locally first before pushing out?
I guess it also depends on what you're building. I mostly work on stuff that isn't used by a consumer. It's going to run as a service on a machine or network of machines far more powerful than my PC. I imagine most people developing for consumers though are building websites or mobile apps. Mobile apps should be tested on a variety of devices, and it would take a pretty intense website to start bogging down most people's machines I imagine.
> it would take a pretty intense website to start bogging down
> most people's machines I imagine.
Not necessarily. On an internal website we found that IE8 took 8 seconds (!) to run a jQuery/jQueryUI command to style all of the buttons on a page. Granted there where more than a couple of buttons, but the point is that DOM manipulation can be a time sink if you don't manage it right. (By contrast the same command took ~1s in Chrome and Firefox, which still isn't insignificant).
An extra second of page load isn't as noticable as 8 extra seconds of page load. Especially on older systems where you might be used to some amount of delay (but not 8 seconds of delay on internal apps).
I guess it depends on what your baseline is. For most of my pageviews, an additional second would double the load time. On an internal app/site an additional few seconds usually isn't worth the trouble, but externally can make a real difference: http://www.zdnet.com/blog/btl/googles-marissa-mayer-speed-wi...
I suspect that builds, or big IDEs, are the biggest performance sinks for devs. If I'm doing non-.NET stack development, I'm pretty sure I could get by with 1 GB of RAM, and probably less if I don't need to have a database instance running on my workstation.
The main thing is to have a process more than a single particular setup. I've tried all sorts of home-built systems, including water cooled, phase change, etc... The conclusion I've reached is that upgrading every 6 months is more important than trying to get the fastest possible system at one point of time. The time and energy I used to spend optimizing my hardware I now spend making sure it's easy to swap out my current machine for a new machine with the least downtime.
Not so much from the perspective of constant hardware upgrades, but from the perspective of being able to translocate and go (and, presumably, also back up an entire environment as easily).
I met Mitch recently at Noisebridge and we discussed the idea of a Linux LiveUSB dev environment for people starting out, but not familiar enough with linux to self set-up.
"I don't know why Apple won't make something in between the Air and an iPhone."
Well that would be the iPad. This is from 2009, I wonder what pg thinks of the iPad regarding this quote. Really makes you notice how fast things change in a year nowadays.
I agree: the site accomplished its only stated goal. As someone who was approached by usesthis to be on their site (just last week: I have not said "no" yet) I can say that the guy who runs it specifically seems to have not cared much about how to optimize for generating insight into the /why/ behind what people are using.
Personally, however, I found that quite disappointing. :( Reading through the list of people and what they had to say I'm thinking, "man, I wonder if that was a random choice, or something that person spent a lot of time thinking about". Maybe I'm weird, but just knowing that someone who's work I admire has a Compaq Concerto isn't interesting if I don't understand the tradeoffs that went into its purchase.
Another complaint I have is that the site attempts what I believe to be a strong and artificial separation between hardware and software, and even requires a fixed ordering: hardware first, software second.
In my experience, software and hardware are really two different physical manifestations of the same underlying concepts (maybe I've spent way too much time on projects that involved writing software to accomplish the same goal as a previous generation's fixed function hardware), and end up having dependency relationships that criss-cross the supposed boundary.
As an example, many people interviewed on the site mentioned that they own a MacBook only because they were recently doing a lot of iPhone development and needed to use Apple's iOS SDK. However, it led to awkward phrasing to do that in the hardware section (before they had a chance to go more in depth into the software later), so the comment is stilted at the top and then short changed later (as it was already mentioned earlier).
It is also quite common, however, for people to first look at physical device properties like form factor, screen size, or input subsystem, leading them to make decisions that sound like "once I decided I had to have a netbook, I started evaluating whether to run Debian or Windows 7: either would be fine as I really only needed to run Chrome and a terminal".
I therefore look at this website as a giant list of missed opportunities to learn something really interesting and important about a number of really awesome people, and while I again agree: the site accomplished its stated goal, I think it is appropriate to judge a site on whether the underlying concept was appropriate in addition to whether they succeeded at their implementation.
Judging by the wide range of profiles, I'd say the format is flexible enough to address all of your concerns without changing a thing. Some of the featured users focus only on what they use to get work done, while others come off as proud technohoarders. Some choose to be explicit about their philosophy, while others let the setup speak for itself. I don't really favor one approach over the other, but I always find the choice of what to include or not include fascinating. In any case, it's one of the few sites where I read every article, so I think they're doing something right.
Can you please provide more detail? I provided an explicit issue--"hardware first, software second"--along with and a specific very common example (iOS SDK) and how I felt that ordering choice (inflexible: I checked with the site's editor) limited the ability for many people to be able to explain that "why" in an effective manner. I would hope you'd be able to show me some counter examples, or show me how I'm simply missing some obvious reinterpretation that would allow me to look past this bug.
The specific hardware and software I use to do my work is something I have spent a lot of explicit time thinking about, both alone and with the people I work with (we have spent so much time bickering over what the absolute best keyboard in the world is it would sicken many people), and I simply do not see a way to make these stories insightful or compelling with the constraint of explicitly discussing hardware first and software second. :(
A final comment: I did not say that they aren't doing anything right. I stated quite explicitly that I "personally" felt there were other interesting goals that were being left aside, and I will say that apparently some other people agree with me or we would not have seen the couple comments in this thread that led me to decide to post something to expand on the thought. Please do not assume that people expressing their personal opinions about something implies that that they believe that thing has no value to anyone.
Depending on your usage, a terminal isn't necessarily minimalistic, apart from the visual side. I've known a lot of people who live in Excel, who could be deemed more minimalistic than a lot of Unix users, using umpteen command-line utilities, a few ad-hoc programming and expression languages…
(Not saying that pg does it that way. For all I know, he might just use vi and mzscheme after all)
In these interviews I rarely come across a piece of software or hardward that isn't accessible from an economic standpoint. Almost always run of the mill stuff. A nice reminder to a guy like me that can get caught up in gear/books/sites, etc. It's not the gear it's 'that' and 'how' you use it.
I just finished reading "Coders at Work". An I 'used' this would make an interesting site as well. Especially if there were pics of some of those old machines. I borrowed a unix book from a friend and he had been using an old punchcard as a bookmark. Neat little find.
edit: Look, I'm not dissing the man. His eyes are angelically beautiful. I must admit, probably not an insight into what tools he uses, but I just hadn't noticed before.
A pet-peeve of mine. I don't want new vim users trying "vi" and having an awful experience ("how do I undo more than once??"), assuming it's not an alias for vim.
I read the lack of mentioning source control this way: it is just assumed that they use some sort of version control system, and for many of them, they are content to use whatever is used in their working environment.
Oxymoron, fellow weekend warrior. Butter must be included. However, it is not an element in the periodic table of deliciousness--because it is so fundamental it is the up-quark of baking.
Check out the Setup's "Community" page, linked from the top: http://usesthis.com/community/ - it includes a list of links to people who have answered the questions on their own blogs.
Paul Graham has a screen smaller then mine and uses two software I didn't know existed which I will now check out, that's everything vaguely "interesting" I found in this.
I don't understand.
Am I not supposed to say that I found this article pointless and everyone is upvoting it just because it has "Paul Graham" in the title?
Sometimes HN really surprises me.
> I'd like it if the HD display was bigger and much lighter and didn't have such a massive octopus of cables attached to it.
Looks like Apple agreed. The more recent Apple displays are bigger and have mostly eliminated the “octopus of cables” (when LightPeak hits this should even further improve), and the new Air is substantially smaller.