I was wondering that as well, but it’s likely a cost/performance tradeoff rather than lack of functionality on macOS.
My preliminary speed tests were fairly slow on my Macbook. However, once I deployed the app to an actual iPhone the speed of OCR was extremely promising (possibly due to the Vision framework using the GPU). I was then able to perform extremely accurate OCR on thousands of images in no time at all, even on the budget iPhone models like the 2nd gen SE.
My preliminary speed tests were fairly slow on my Macbook. However, once I deployed the app to an actual iPhone the speed of OCR was extremely promising (possibly due to the Vision framework using the GPU).
Not clear why the MacBook was so much slower, however!
Linux productivity software is fine, but there are rough edges for the power user
Interesting. For me, it is the exact opposite. I used Linux as my main desktop from 1994 to 2007. In 2007 I switched to macOS as my main desktop OS. The last few years I have slowly been moving back to Linux as my main desktop. After a short adventure with the M1 MacBook Air, I decided that the Mac ecosystem is definitely on a trajectory that I don’t like anymore, and returned the M1 Mac and sold my last Intel Mac.
What I like about the Linux desktop is that it is so much beter for the power user. Better profiling (perf), better observability (eBPF, yes there is DTrace, but the eBPF ecosystem is so much more exiting), better customizability, better development tools, better package managers, native containers, etc.
What I dislike is the lack of high-quality productivity software (and since when are e-mail clients and calendaring apps not productivity software?). Sure, we have LibreOffice, GIMP, Inkscape, Scribus. But I would love to be able to use Microsoft Office, the Affinity Suite, OmniGraffle, etc. Many productivity apps are just too far ahead of open counterparts. So, even though I moved away from the Mac, I now have to keep a Windows partition around to run Microsoft Office.
FWIW DTrace is completely broken out of the box on OS X. Apparently it’s fixable but you have to disable SIP. This makes me extremely sad.
If I find myself in the position of needing to understand something very weird that a piece of software is doing, if it’s portable across unices then it’s sometimes easiest to run it in a Linux VM since things like strace actually work there. :(
When something isn’t working on my Linux workstation I similarly take it over to an illumos VM where truss and DTrace are available haha, so I know the feeling.
Not going to lie, I think I’d be tempted to do the same if I did know dtrace (and to a lesser extent illumos, but eh I assume it’s well made enough to not be difficult).
I don’t because of a lack of familiarity with the tools.
When I’m using Linux I still miss ftrace from BSD. strace is great except that the probe effect is so bad.
What I dislike is the lack of high-quality productivity software (and since when are e-mail clients and calendaring apps not productivity software?).
This is something I’ve noticed as well. Linux developers prioritize things that they’re interested in (perf, eBPF, DTrace), meaning that the stuff that they don’t care about (productivity software, especially email) lags behind Windows/macOS by years (if not decades) - but commercial developers won’t port their stuff to Linux, because (in addition to the market being small) Linux users have been habituated to not pay for software!
I’m inclined to say that this is a product of the FSF’s rabid anti-commercial-software propaganda, but maybe there’s another reason.
I think it’s simply more to do with power structures – like, the Mac has a good email client, because someone at Apple pays people to make it, and more importantly, decides what it’s going to do. Free software has tended to operate without a singular product vision, so in order to make progress, the people involved seem to try and clone what they’re familiar with.
but commercial developers won’t port their stuff to Linux, because (in addition to the market being small) Linux users have been habituated to not pay for software!
I’m inclined to say that this is a product of the FSF’s rabid anti-commercial-software propaganda, but maybe there’s another reason.
I thought we debunked this theory when Steam came to Linux.
The truth is that it’s just the old chicken and egg problem about small market share and market share won’t grow until peoples’ favorite software is available.
I thought we debunked this theory when Steam came to Linux.
Computer users have completely different expectations for games and productivity software, and are far more willing to pay for the former across every platform. Steam coming to Linux did not “debunk” that theory at all.
The truth is that it’s just the old chicken and egg problem about small market share and market share won’t grow until peoples’ favorite software is available.
Please don’t say things like “the truth is” without any irrefutable substantiating evidence.
Meanwhile, that’s your spin on things. I would then ask you why macOS productivity software is so much better-developed than Linux, when the former already has a tiny market share compared to Windows.
Computer users have completely different expectations for games and productivity software, and are far more willing to pay for the former across every platform. Steam coming to Linux did not “debunk” that theory at all.
I agree that computer users are much more willing to pay for games across every platform. I’m not sure how that debunks my debunking. I’m asserting that Linux users are just as willing to pay for Steam games as people on any other platform (though we obviously don’t have hard data on that). If that’s true (or even close to true), then I think that does debunk the argument that they aren’t buying other software because of “being habituated to not paying for software”. Linux forums EXPLODED when Steam came to Linux. People were losing their minds to throw money at Valve. I would think that the anti-spending habituation would take time to wear off, but it did not seem to in the internet circles that I frequent (and the small handful of people I knew in meatspace that ran Linux also buy/bought Steam games).
Along similar lines, I don’t think I know any Windows users that are buying Microsoft Office or Adobe PhotoShop for personal use, either. It seems to me that the willingness to pay for entertainment and not for productivity software is more or less universal…
Please don’t say things like “the truth is” without any irrefutable substantiating evidence.
Fair enough. I suppose I could’ve been more clear in indicating that this was a statement of conjecture. (But irrefutable? That’s a high bar for a semi-casual discussion about human behavior, no? ;) )
Meanwhile, that’s your spin on things. I would then ask you why macOS productivity software is so much better-developed than Linux, when the former already has a tiny market share compared to Windows.
Can you give examples? I ask that genuinely, just so I can understand what we’re talking about. I do mostly run Linux on my personal machines and have a Mac for work. But I’m a software dev, so I really don’t do things like PhotoShop, Microsoft Word, etc. So I just don’t know what we’re talking about. I know that when I was in science, the Linux machines all had MATLAB and Mathematica installed on them. Those are pricey.
But, I can offer some hypotheses other than Linux users not liking to pay for things other than games:
Apple’s market share is still a gazillion times bigger than desktop Linux’s. If there’s a threshold for when it would become profitable to support a platform, it’s entirely conceivable that Macs have crossed that threshold and Linux has not.
Related to the above point, there isn’t even one Linux platform. It’s probably harder to develop (and distribute) for Linux than either of the other major platforms. These days it is probably easier with Snap and Flatpack and whatnot, but historically this has not been true. So the cost would be higher to port to Linux than to port to Mac, for example.
The “chicken and egg” scenario that I mentioned before.
Similar to the “chicken and egg”, but maybe people who use Linux don’t need the same kind of productivity software for the ways they are productive with their machines. For example, I know several people who run Linux and pay for IntelliJ licenses. So, even if $TOTAL_NUMBER_OF_LINUX users was large enough, there still just aren’t enough of them who want PhotoShop functionality to make it worth it.
the stuff that they don’t care about (productivity software, especially email) lags behind Windows/macOS by years (if not decades)
Another thing may be is that if you have been using Linux for most of your life like me, I have no idea what this productivity software has to offer me. E.g. I can use Email, send and recieve messages, send plaintext messages, my client understands the headers, etc. That is what I imagine Email to be. What would an Email-Salesman tell me to change my mind, and switch to some propriatory client that is entierly different?
the stuff that they don’t care about (productivity software, especially email)
I think it’s a little trickier. A lot of Linux developers do care about some kinds of productivity software, but many of the people in this category are also not very excited about GUI apps, so a lot of the actively maintained and innovative software is either terminal software or even an emacs package, making it a bit inaccessible to people looking for Windows/macOS style apps.
For example I use mu4e as my email client for a few years now. It’s actively developed (multiple yearly releases, regular new features, quick bugfixes, etc.). In my opinion has one of the best UIs out there for a power user, but it is almost certainly not what someone migrating from Mail.app wants.
People who don’t want to pay can do this via piracy as well, and also take advantage of anybetter software that may not have a counterpart on Linux. Also, free software is not about not paying. If the values of free software collide with the commercial intentions of software developers does not mean that the former is propaganda against the latter.
People who don’t want to pay can do this via piracy as well
Piracy happens on Linux too - so now you have two reasons why your application won’t make as much money. (well, three, but the third (tiny market share) isn’t relevant)
If the values of free software collide with the commercial intentions of software developers does not mean that the former is propaganda against the latter.
Beyond the general “values of open-source”, the FSF in particular (which I specifically mentioned) has engaged in anti-commercial-software propaganda.
I switch to my laptop, which I have hooked up to the same screen. I checked the CrossOver database, from from which I inferred that e.g. recent Office versions do not really work. Running Windows in a VM would definitely be possible, but haven’t set it up yet.
I just had a look and yes the star ratings have been dropping for the recent versions. It’s a bit of a shame because I remember using it years ago and it felt surprisingly ‘native’ - even in how snappy it was.
Yeah, I also used CrossOver Office in the early ’00s and it was amazing. I think one of the issues is that back then, productivity applications were still pretty much shrink-wrap, static pieces of software. Nowadays, things like Office (especially with Office 365) are constantly updated and therefore a moving target.
I should try the VM route again, but the last time it was not that great. I really dislike VirtualBox and AFAIR graphics acceleration was not really great with SPICE. VMWare probably doesn’t work well with NixOS and Fedora.
I feel exactly the same way, although I don’t use most of the mentioned apps (Photoshop, Calendar, Mail, etc)
After discovering i3 and investing way too much time customizing my shell experience I’m ready to ditch OSX for good. My maxed out ThinkPad x1 Gen8 screams relative to my maxed out 16” MBP (granted it has all the performance sucking endpoint protection spyware installed). That said, part of my day job is iOS build tools so it won’t be so easy to fully get away :)
We have several “services” (not the micro-variety) that use a shared database. Each service has its own user with limited access to the tables it needs, but more importantly it lets us use a tool like PgHero or PgAnalyze to look at database usage patterns on a per-service basis. This has helped a lot with identifying bad queries, application bugs, etc.
I wonder if there’s a reason they didn’t use a fleet macOS machines running multiple iOS simulators.
I was wondering that as well, but it’s likely a cost/performance tradeoff rather than lack of functionality on macOS.
I wonder if his MacBook is an Intel one or an Apple Silicon one. The latter has an architecture closer to the iPhone’s.
Starting with a few $40 iPhones SE 2 probably factored in to the calculation.
Or run the Vision APIs on macOS, when it’s documented as supported….
I’m aware of ocrit, a command-line tool that uses Apple’s Vision for OCR.
The article says:
Not clear why the MacBook was so much slower, however!
Wasn’t he simulating iPhone in XCode when testing on his macbook, though?
Interesting. For me, it is the exact opposite. I used Linux as my main desktop from 1994 to 2007. In 2007 I switched to macOS as my main desktop OS. The last few years I have slowly been moving back to Linux as my main desktop. After a short adventure with the M1 MacBook Air, I decided that the Mac ecosystem is definitely on a trajectory that I don’t like anymore, and returned the M1 Mac and sold my last Intel Mac.
What I like about the Linux desktop is that it is so much beter for the power user. Better profiling (
perf), better observability (eBPF, yes there is DTrace, but the eBPF ecosystem is so much more exiting), better customizability, better development tools, better package managers, native containers, etc.What I dislike is the lack of high-quality productivity software (and since when are e-mail clients and calendaring apps not productivity software?). Sure, we have LibreOffice, GIMP, Inkscape, Scribus. But I would love to be able to use Microsoft Office, the Affinity Suite, OmniGraffle, etc. Many productivity apps are just too far ahead of open counterparts. So, even though I moved away from the Mac, I now have to keep a Windows partition around to run Microsoft Office.
FWIW DTrace is completely broken out of the box on OS X. Apparently it’s fixable but you have to disable SIP. This makes me extremely sad.
If I find myself in the position of needing to understand something very weird that a piece of software is doing, if it’s portable across unices then it’s sometimes easiest to run it in a Linux VM since things like strace actually work there. :(
When something isn’t working on my Linux workstation I similarly take it over to an illumos VM where
trussand DTrace are available haha, so I know the feeling.Not going to lie, I think I’d be tempted to do the same if I did know dtrace (and to a lesser extent illumos, but eh I assume it’s well made enough to not be difficult).
I don’t because of a lack of familiarity with the tools.
When I’m using Linux I still miss ftrace from BSD. strace is great except that the probe effect is so bad.
This is something I’ve noticed as well. Linux developers prioritize things that they’re interested in (perf, eBPF, DTrace), meaning that the stuff that they don’t care about (productivity software, especially email) lags behind Windows/macOS by years (if not decades) - but commercial developers won’t port their stuff to Linux, because (in addition to the market being small) Linux users have been habituated to not pay for software!
I’m inclined to say that this is a product of the FSF’s rabid anti-commercial-software propaganda, but maybe there’s another reason.
I think it’s simply more to do with power structures – like, the Mac has a good email client, because someone at Apple pays people to make it, and more importantly, decides what it’s going to do. Free software has tended to operate without a singular product vision, so in order to make progress, the people involved seem to try and clone what they’re familiar with.
I thought we debunked this theory when Steam came to Linux.
The truth is that it’s just the old chicken and egg problem about small market share and market share won’t grow until peoples’ favorite software is available.
Also because Valve had two really good reasons to seek out wider OS support: the Windows and Mac App Stores.
Computer users have completely different expectations for games and productivity software, and are far more willing to pay for the former across every platform. Steam coming to Linux did not “debunk” that theory at all.
Please don’t say things like “the truth is” without any irrefutable substantiating evidence.
Meanwhile, that’s your spin on things. I would then ask you why macOS productivity software is so much better-developed than Linux, when the former already has a tiny market share compared to Windows.
I agree that computer users are much more willing to pay for games across every platform. I’m not sure how that debunks my debunking. I’m asserting that Linux users are just as willing to pay for Steam games as people on any other platform (though we obviously don’t have hard data on that). If that’s true (or even close to true), then I think that does debunk the argument that they aren’t buying other software because of “being habituated to not paying for software”. Linux forums EXPLODED when Steam came to Linux. People were losing their minds to throw money at Valve. I would think that the anti-spending habituation would take time to wear off, but it did not seem to in the internet circles that I frequent (and the small handful of people I knew in meatspace that ran Linux also buy/bought Steam games).
Along similar lines, I don’t think I know any Windows users that are buying Microsoft Office or Adobe PhotoShop for personal use, either. It seems to me that the willingness to pay for entertainment and not for productivity software is more or less universal…
Fair enough. I suppose I could’ve been more clear in indicating that this was a statement of conjecture. (But irrefutable? That’s a high bar for a semi-casual discussion about human behavior, no? ;) )
Can you give examples? I ask that genuinely, just so I can understand what we’re talking about. I do mostly run Linux on my personal machines and have a Mac for work. But I’m a software dev, so I really don’t do things like PhotoShop, Microsoft Word, etc. So I just don’t know what we’re talking about. I know that when I was in science, the Linux machines all had MATLAB and Mathematica installed on them. Those are pricey.
But, I can offer some hypotheses other than Linux users not liking to pay for things other than games:
Another thing may be is that if you have been using Linux for most of your life like me, I have no idea what this productivity software has to offer me. E.g. I can use Email, send and recieve messages, send plaintext messages, my client understands the headers, etc. That is what I imagine Email to be. What would an Email-Salesman tell me to change my mind, and switch to some propriatory client that is entierly different?
I think it’s a little trickier. A lot of Linux developers do care about some kinds of productivity software, but many of the people in this category are also not very excited about GUI apps, so a lot of the actively maintained and innovative software is either terminal software or even an emacs package, making it a bit inaccessible to people looking for Windows/macOS style apps.
For example I use mu4e as my email client for a few years now. It’s actively developed (multiple yearly releases, regular new features, quick bugfixes, etc.). In my opinion has one of the best UIs out there for a power user, but it is almost certainly not what someone migrating from Mail.app wants.
People who don’t want to pay can do this via piracy as well, and also take advantage of anybetter software that may not have a counterpart on Linux. Also, free software is not about not paying. If the values of free software collide with the commercial intentions of software developers does not mean that the former is propaganda against the latter.
Piracy happens on Linux too - so now you have two reasons why your application won’t make as much money. (well, three, but the third (tiny market share) isn’t relevant)
Beyond the general “values of open-source”, the FSF in particular (which I specifically mentioned) has engaged in anti-commercial-software propaganda.
What didn’t you like about it? I personally love my M1 MacBook Air, but my preferences certainly aren’t universal truths.
Do you reboot into Windows to run Office, use virtualisation software, use CrossOver, or something else?
I switch to my laptop, which I have hooked up to the same screen. I checked the CrossOver database, from from which I inferred that e.g. recent Office versions do not really work. Running Windows in a VM would definitely be possible, but haven’t set it up yet.
I just had a look and yes the star ratings have been dropping for the recent versions. It’s a bit of a shame because I remember using it years ago and it felt surprisingly ‘native’ - even in how snappy it was.
Yeah, I also used CrossOver Office in the early ’00s and it was amazing. I think one of the issues is that back then, productivity applications were still pretty much shrink-wrap, static pieces of software. Nowadays, things like Office (especially with Office 365) are constantly updated and therefore a moving target.
I should try the VM route again, but the last time it was not that great. I really dislike VirtualBox and AFAIR graphics acceleration was not really great with SPICE. VMWare probably doesn’t work well with NixOS and Fedora.
I feel exactly the same way, although I don’t use most of the mentioned apps (Photoshop, Calendar, Mail, etc)
After discovering i3 and investing way too much time customizing my shell experience I’m ready to ditch OSX for good. My maxed out ThinkPad x1 Gen8 screams relative to my maxed out 16” MBP (granted it has all the performance sucking endpoint protection spyware installed). That said, part of my day job is iOS build tools so it won’t be so easy to fully get away :)
We have several “services” (not the micro-variety) that use a shared database. Each service has its own user with limited access to the tables it needs, but more importantly it lets us use a tool like PgHero or PgAnalyze to look at database usage patterns on a per-service basis. This has helped a lot with identifying bad queries, application bugs, etc.