Yeah he's lamenting how poorly the external developers are being treated but seems like they are largely being treated the same as Apple treats its own employees
I found it sometimes frustrating as well. But my experience was that it depended on the engineer/team/screener the Radar went to. Obviously if it was a bug I found, I much preferred to talk to the engineer in person who maintained the code in question.
There were plenty of teams that were very good about jumping on bug reports and getting right back to me. In fact, a decade or so ago at Apple this was more or less the rule.
My sense though is that engineering became overwhelmed with bugs especially as Apple itself became as successful as it has become.
"In the old days" a smaller team would have one of the engineers on the team screen Radars — maybe rotating to a new engineer every week or so. Engineers could quickly mark a bug as a duplicate of another (recognizing a familiar backtrace or symptoms) or might be able to guess at a diagnosis and send it straight to the engineer in charge of the (likely) relevant code.
A few teams I worked on were very good about daily "BRB"s (bug review board meetings). With a number of engineers in attendance, again, bugs were quickly dispatched to the right component, assigned to right engineer.
Again though, as Apple grew, I saw bug screeners on many teams more or less automatically flipping Radars back to the Originator if there was not a litany of documentation attached to the Radar. What build of the OS, what hardware, a full sample, system diagnosis.... It was frustrating when it was an obvious bug — maybe even a UI bug — that would take someone all of 30 seconds to repro. I somewhat sympathize though as traffic no doubt began to become a firehose of bug reports.
It seemed backwards to me though to put such a burden on the person who is simply trying to report the bug, trying to help you improve the software. Often in screening Radars the description alone was enough for me to know exactly where in the code the problem was. I didn't mind my time being spent at least scanning the Radar title and description. If it was not clear to me at that point, sure, flip it back for more info. The seeming rise of "auto-reject" is what began to irk me.
Yes, if there's a stack trace attached. That is kind of cool though — they also bucket them by OS release so you can see quickly when a bug first appeared (or, well, a similar stack trace appeared anyway). Also they rank them on frequency so it is easy to find your component's "Top Crasher".
Ideally, daily BRBs (kept to an hour) with engineers attending.
If it turns out there are too many Radars to keep the BRBs to an hour then there are issues elsewhere.
At the same time, I didn't mention it above, engineers need to be allowed time in the product schedule to fix these bugs. As probably everyone knows, when a bug is kicked down the road once (because of scheduling constraints: hey, we gotta ship the software at some point!) it becomes easier still to kick it down the road when the next release dat rolls around (we lived with the bug this long...).
I think engineers and users alike would love to see more bug--only releases of MacOS and iOS. People still talk about how wonderful SnowLeopard was simply because it was a bug-fix only (mostly?) release.
I think engineers and users alike would love to see more bug--only
releases of MacOS and iOS. People still talk about how wonderful Snow
Leopard was simply because it was a bug-fix only (mostly?) release.
Apple's dot zero OSX releases have always been pretty bad, so yes Snow Leopard and the last iteration of Tiger look great by comparison. As an outsider I don't have fond memories of filing bugs with Apple in that era. The general opacity combined with corporate disinterest discouraged me to the point where it's been years since I've even thought about filing a bug.
I dread new iOS releases. Sure there are usually unwanted changes (e.g. text selection in iOS 13), but the bugs... iOS 16 completely trashed whatever database Siri uses to map text to speech, so using Siri to play a song is usually a miss. Most recently it's decided the name Ben is really the name "Mouse Face". That's left a fair bit of egg on my face as I interact with those two people in two very different contexts. Sure, I could file a bug, but it's just easier to stop using Siri (which was never great in the first place).
With Tiger I was trying to integrate LDAP (AD masquerading as OpenDirectory), and it turned out that OSX would just assign GID 0 to any LDAP backed user. To me that seems pretty fucking straightforward, obviously pretty severe. That bug report went nowhere (not even an acknowledgement) and if memory serves Leopard went out with the same issue.
I also had a polycarbonate MacBook for personal use around the same time. I ended up filling out the RAM slots and discovered that after a certain amount of uptime that playing DVDs would eventually cause a kernel panic. That bug report actually did get a response, but only one. I think the computer itself died before that got fixed.
As someone who went on to dick around with CalDAV and CardDAV, there were plenty of issues to be filed but it was never really worth the effort. But as with everything else, as bad as Apple is they're usually better than the competition. That's how I ended up replacing my recently deceased Intel MBP with a new ARM one.
So long as market share holds I don't think Apple's got much incentive to reform. Although given that Tim Apple seems hell bent on pulling a Gil Amelio (my new MacBook SuperproairMax 15.323456" with the all new M-Centris really truly pro max upper bound superlative CPU sure looks great on the shelf) who knows how much longer that'll hold.
I was on the team that owned CalDAV and CardDAV in approximately that era, as a bug screener and with zero decision making power.
The attitude was basically, if it doesn't affect us personally working inside Apple, we don't care. If our personal workflow and servers doesn't use those features (which I guess is the case for GID), it doesn't matter.
We don't know what Exchange is or why anyone would use a Microsoft product. What are they, idiots? We don't really care about Gmail. Why is their server not responding according to RFC Whatever? Google must be idiots.
Etc.
There were a very few number of developers, most were wasting their time making fake leather stitching for the UI because Jobs demanded it. Unit testing was quite poor. Code review was nonexistent.
I should probably clarify: the DVD bug was related to playing DVDs on a Mac with Intel graphics and 4 GB of RAM. Certainly it's an odd edge case but it spoke volumes about the quality control with the drivers. Maxing out the RAM on one of those can't have been such a rare thing. Certainly I can wrap my head around why calendaring and enterprise features didn't get much love, but video drivers on the mass market product? Jesus.
With the calendaring stuff: I wrote a CardDAV server that's probably still lurking around on Github somewhere. The workarounds I had to implement for MacOS and iOS clients were beyond pale. For instance stuff would just get cached indefinitely and trailing slashes caused OSX to have a conniption fit. If you told me that the client implementations just hardcoded a bunch of stuff so it worked for Steve Jobs (and that nobody else used it) I'd believe you.
Eventually I did a brief stint at a consultant that was brought on to do some infra stuff for Apple, and that didn't do much to improve my impression. We got a bunch of servers from the basement with expired iLO licenses, and boy were we lucky to get that much.
The attitude was basically, if it doesn't affect us personally working inside Apple,
we don't care. If our personal workflow and servers doesn't use those features (which
I guess is the case for GID), it doesn't matter.
Honestly I'm not sure how to respond to this. If memory serves there was no way to uncouple group mapping, so this was part and parcel of Apple's push into the enterprise space (and in this case AD was behaving as an OpenDirectory server would've). Which is to say, their failure in that space was entirely a self-fulfilling prophecy. The more frustrating part was that an obvious security flaw was treated with such disdain, which encouraged me to stop caring about filing bug report with Apple (and it's lead me to keep the bluetooth modem off almost exclusively on my iPhone).
That gig was actually mostly staffed with Apple alumni so there were 1st gen iPhones all around. Even the less esoteric stuff was simply not good. Obviously voice quality was shit (at least in San Francisco) compared to the CDMA feature phones of the day. One of the coworkers there worked on the mobile Mail.app before bailing for startup life. Autocomplete was a mess. Honestly, I rather enjoyed rubbing some salt into that wound.
What amazes me is that fifteen years on and Microsoft is still somehow worse (what with Windows being adware and all).
>People still talk about how wonderful SnowLeopard was simply because it was a bug-fix only (mostly?) release.
I think that was part marketing, snow leopard brought pretty sweeping architectural/internal changes: ARC, GCD, 64-bit kernel, so it definitely wasn't a pure bugfix release.
When I was there on the BRBs I often dreamed of actually commissioning a neon *NMOS* sign that I would flash by pressing a button for nearly every bug that came up for discussion.
