Oh that’s a great test case. I’ve been annoyed by the speed issue too. For a while I had json and Md files bound to Zed because it was faster with highlighting than anything else.
But even that takes seconds to start on my M1 Max from cold. So then I just wrote (avec Claude) a viewer app and it turns out computers are so fast you can not only open multi GB JSON files instantly you can even pretty print them so fast the user won’t lose attention.
But I just generated those. Moby Dick is a great real-world test case.
One can keep it open without a window on MacOS I think but in the end the vibe-coded solution outperforms it on this specific task (which is not its primary task) at a few orders of magnitude. It is perhaps on the frontier for speed-features but I don’t need the features so when I move backwards along the curve I can extract more speed myself.
That reminded me of how fast/instant Winamp loads the huge directory (recursively), while other players struggle. Scrolling reveals that files are being loaded lazily (obviously), but the most interesting part is that Quick search and Jump to file works immediately.
> both (IMHO) are much longer than they need to be
I haven’t read moby dick but i dropped 20’000 leagues under the see a bit past half the book because of this. At some point the author spent pages and pages and pages describing the environment under the see, often repeating himself.
I’ll get back to that book at some point but yes, it’s longer than it needs to be.
I used war and peace to explain to a CEO how wasteful a mobile API was; twelve distinct items of data to display but we sent JSON larger than war and peace.
I used to use this as my example for how big a megabyte actually is. I think the entire text is 4 MB uncompressed (a nagging voice in my head says 16 MB, but I think that's just the residual part of me that never really believed how big a MB actually is).
A bit amusing. I took Anna Karenina off Gutenberg and used that as test data for some of my flashing algorithms. I called it the "Anna test". I could have used random data, but where's the fun in that? Besides, during dev, structured text showed the kind of error I got much better than random data would have.
not to pile on the particular software but the example just stuck in my memory, two years ago or so I tried out Logseq for note taking, and I still remember that it put a five page file (not even Moby Dick) into 'read only' mode because apparently at about 1k characters or a few hundred lines of text the app couldn't handle the performance impact, stumbling across discussions like this[1]
With the quasi supercomputers we have, that somehow apps that exist to edit and display text crap themselves on ordinary <1mb files is just weird. There should be no trade-offs.
Oh that’s a great test case. I’ve been annoyed by the speed issue too. For a while I had json and Md files bound to Zed because it was faster with highlighting than anything else.
But even that takes seconds to start on my M1 Max from cold. So then I just wrote (avec Claude) a viewer app and it turns out computers are so fast you can not only open multi GB JSON files instantly you can even pretty print them so fast the user won’t lose attention.
But I just generated those. Moby Dick is a great real-world test case.
This is why you use Emacs in Daemon mode for instant starts. Does zed perhaps offer something similar?
One can keep it open without a window on MacOS I think but in the end the vibe-coded solution outperforms it on this specific task (which is not its primary task) at a few orders of magnitude. It is perhaps on the frontier for speed-features but I don’t need the features so when I move backwards along the curve I can extract more speed myself.
That reminded me of how fast/instant Winamp loads the huge directory (recursively), while other players struggle. Scrolling reveals that files are being loaded lazily (obviously), but the most interesting part is that Quick search and Jump to file works immediately.
This reminds me of ‘The Jules Test’. Popularised (hyper locally) by my friend Jules.
When renting a flat, simply head straight to the bathroom and flush the toilet. If the toilet flush is good, the flat is fine.
You can use "20,000 leagues under the sea" instead of Moby Dick.
Both were written at the same time, both deal with ocean adventures, and both (IMHO) are much longer than they need to be.
Then you can have multiple Jules tests, depending on occasion!
Moby Dick had a much worse ratio of "size of good novel trying to get out / actual size", in my opinion.
> both (IMHO) are much longer than they need to be
I haven’t read moby dick but i dropped 20’000 leagues under the see a bit past half the book because of this. At some point the author spent pages and pages and pages describing the environment under the see, often repeating himself.
I’ll get back to that book at some point but yes, it’s longer than it needs to be.
Turn on the sink before you flush so a sizeable drop in header pressure becomes obvious.
And here I thought this was going to be, “do a push up every time you have to look up a word in the dictionary while reading Moby Dick.”
Or how to stay fit like an 1800s whaler. Where am I going to find a hook and a 200 pound chunk of whale blubber though?
Find your nearest Discord Mod Meetup.
Could substitute a tractor tire, maybe? Would certainly smell better long term…
I use "War and Peace" for this.
I used war and peace to explain to a CEO how wasteful a mobile API was; twelve distinct items of data to display but we sent JSON larger than war and peace.
I used to use this as my example for how big a megabyte actually is. I think the entire text is 4 MB uncompressed (a nagging voice in my head says 16 MB, but I think that's just the residual part of me that never really believed how big a MB actually is).
Good idea, I might use Crime & Punishment!
A bit amusing. I took Anna Karenina off Gutenberg and used that as test data for some of my flashing algorithms. I called it the "Anna test". I could have used random data, but where's the fun in that? Besides, during dev, structured text showed the kind of error I got much better than random data would have.
did jesse write this? i presume so but maybe hog bay has an anonymous forum
I did! And I'm still using it to test Bike 2.0. Will will be release about 2 years ago. :\ Well soon anyway!
not to pile on the particular software but the example just stuck in my memory, two years ago or so I tried out Logseq for note taking, and I still remember that it put a five page file (not even Moby Dick) into 'read only' mode because apparently at about 1k characters or a few hundred lines of text the app couldn't handle the performance impact, stumbling across discussions like this[1]
With the quasi supercomputers we have, that somehow apps that exist to edit and display text crap themselves on ordinary <1mb files is just weird. There should be no trade-offs.
[1] https://discuss.logseq.com/t/logseq-unusable-for-long-form-p...
(2022)
You had me at "dick workout".
And to Moby music! What's not to love?