Magnes blog

Musings from a norwegian entrepreneur, web developer and armchair philosopher.

[magnemg]

How to fix sudden FPS drop in CS:GO (or CS2) on Macbook Pro 2019 with AMD Radeon 560X (and optimise FPS performance)

Posted: 2023-10-04 (settings tested just before and after the CS2 release around this date) CS:GO = Counter-Strike: Global Offensive CS2 = Counter-Strike 2

Have you cleaned the fans inside the casing? Open the Macbook (with a Torx TS4 screwdriver, from a Bits set from your local hardware store), and vacuum the fans inside. They have likely collected dust over the years, which make the Macbook unable to disperse of the heat when it is put under stress (lots of things happening ingame) which causes the CPU to emergency throttle because it becomes too hot.

Other things you may try:

https://www.bootcampdrivers.com/ for better AMD Radeon drivers than the latest. For Macbook Pro 2019 with AMD Radeon Pro 560X then ideally use the February 2022 edition or older of the Adrenalin drivers, like the site Compatibility section says. The closest is the Arenalin March 2022 - Red Gaming Edition drivers, which has been confirmed to work.

This plus the frame rate per second (FPS) optimisation settings below can give you occasionally 300-400 fps in CS:GO ! FPS in CS:GO should not drop to below 70 FPS when in the heat of action, with these settings. For CS2 then you should be able to get around 120 FPS stable, with a drop to around 100 FPS when in the heat of action

Many have experienced FPS to drop by around 30 fps in the heat of action, in CS2 after it was released. But my experience has been that the fps has actually been more stable in CS2 than in CS:GO, with the setup and specs mentioned here.

MSI Afterburner default settings:

Core Clock (Mhz): 1004 Memory clock (Mhz): 1470 (Volt is around 750 mV, temp around 55 deg, on startup, at evening).

(with the below settings, and player contrast and multicore rendering enabled)

This can get >98 fps average in the Workshop map that tests FPS! Without frame drops / lags / stutters!

Use ThrottleStop software! To fix Macbook Pro CPU throttling when it asks for too much power. That is the key, after cleaning the fans. The rest is minor.

Benchmark in-game in CS:GO:

Steam: Right click CS GO -> Workshop -> Search: "FPS Benchmark". Direct link: https://steamcommunity.com/sharedfiles/filedetails/?id=500334237

Subcribe to it. https://steamcommunity.com/sharedfiles/filedetails/?id=500334237

ThrottleStop software:

High Performance enabled

Speed Shift EPP enabled: 60 (or 32; lower will cap the CPU mhz sooner).

Disable Turbo to fix spikes (to disable throttling due to overheating!).

References:

https://www.youtube.com/watch?v=cn2MO4eJo4k

https://www.reddit.com/r/macgaming/comments/gmpqj5/guide_double_your_fps_with_process_lasso/

Process Lasso software:

csgo.exe right click -> Affinity -> Disable SMT (HyperThreading), so that it uses each physical core more effectively (less thread contention, games not so good at multithreading, singlethreading can give 30% better clock performance). -> CPU priority -> Always -> High -> I/O priority -> Always -> High -> Memory priority -> Always -> Normal (which is the highest) -> Power profile -> Max performance overlay

CS:GO Graphics and Advanced Graphics settings:

Normal 4:3 aspect ratio (AMD software should stretch it to screen size; like the professionals do, or try 1280x1024 for most stretch, but maybe less FPS).

800x600 resolution

Fullscreen

Texture Streaming: OFF (can otherwise cause stutter)

Multicore rendering: Doesn't matter (not much diff during tests), but try with ON

Laptop power saving: OFF (will otherwise set fps_max to 48. Plus maybe throttle CPU / GPU??)

CS:GO Launch settings:

Steam: Right click CS GO icon -> General -> Startup settings (then paste this line there): -untrusted -high -tickrate 128 +fps_max 0 -novid -nojoy -fullscreen +mat_disable_fancy_blending 1 +cl_forcepreload 1 -limitvsconst -forcenovsync +r_dynamic 0 -nothreadedsockets

CS:GO autoexec.cfg file in folder "C:\Program Files (x86)\Steam\steamapps\common\Counter-Strike Global Offensive\csgo\cfg" :

See contents of autoexec.cfg below. Commands that will be auto executed when joining a game (or starting CS:GO perhaps).

AMD Radeon Pro software:

Create -> Global graphics -> Frame rate target control (60 fps max, or could be set lower)

Frame rate target control: Disabled (seems to give less stable FPS than controlling this ingame).

Texture filtering quality: Performance

If wanting to cap FPS: Probably best to simply set fps_max to 100 or 120 ingame, since I am not sure that the ingame laptop power setting does not affect anything else but the FPS max (like CPU or GPU throttling or whatever).

MSI Afterburner:

Settings unknown. I didn`t get it to work properly, the settings always reset when I clicked Apply. But others have used it to underclock the GPU.

----

CS GO Launch option settings:

-novid -nojoy -tickrate 128 -nothreadedsockets

Only these and nothing else! According to professional optimization expert. https://youtu.be/ogNoiim9ZwM?t=154

novid = disable intro video when launching nojoy = disable joystick tickrate = when hosting a local server

nothreadedsockets = better hit registration, more: https://www.reddit.com/r/GlobalOffensive/comments/10okxes/hey_guys_its_nartouthere_i_collab_with/j6muylt/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button

---

Put this in autoexec.cfg in

C:\Program Files (x86)\Steam\steamapps\common\Counter-Strike Global Offensive\csgo\cfg :

net_threaded_socket_burst_cap "" // =empty value . Any non-empty -> worse performance

net_threaded_socket_recovery_rate "X.XXXXXX"

X.XXXXXX -> worse than int X (new bug with type converatation?)

net_threaded_socket_recovery_time "60.000000"

ex.: "" worse than "60" |

demo_index "1" cl_join_advertise "2"

rate "786432"

cl_cmdrate "128" cl_updaterate "128" cl_interp_ratio "2" cl_interp "0" cl_lagcompensation "1" cl_predict "1" cl_predictweapons "1"

m_rawinput "1" m_pitch "0.022" m_customaccel "0" m_mouseaccel1 "0" m_mouseaccel2 "0" m_customaccel "0" m_mouseaccel1 "0" m_mouseaccel2 "0" m_mousespeed "0"

fps_max "999" // to see how high you can go, you should be able to have 300-400 fps occasionally ingame if you have AMD Radeon 560X. Change this setting once you get in game to 0. When you have fixed the problem you might want to set it to 120 to have a more stable FPS

fps_max_menu "999" r_dynamic "0" r_drawtracers_firstperson "1" cl_threaded_bone_setup "1"

ui_steam_overlay_notification_position "bottomright" player_nevershow_communityservermessage "1" mm_dedicated_search_maxping "25" // max ping for matchmaking games

r_eyegloss "0" // No Makeup. r_eyemove "0" // No Eye Movement. r_eyeshift_x "0" // No Eye Movement X-Angel. r_eyeshift_y "0" // No Eye Movement Y-Angel. r_eyeshift_z "0" // No Eye Movement Z-Angel. r_eyesize "0" // Smaller Eyes. m_rawinput "1"

clear

[magnemg]

Wordpress Multisite approve new users (admin moderation)

Problem: You want to have an admin moderation tool that allows you to approve new users that register to your Wordpress Multisite. You’ve tried the New User Approve plugin, but realise it doesn’t work for Wordpress Multisite installations. You’ve found the unofficial workaround called the New User Approve MU plugin, but you didn’t get that working either. Maybe because you have Wordpress 4.x installed and that plugin fix came out in 2012. You’ve scoured the web looking for good alternatives, but haven’t found any. Solution: It’s as simple as using the Gravity Forms User Registration Add-on. It will give you what you seek, and you might already be using GravityForms, so it will fit like hand in glove!

[magnemg]

How to reduce iPhoto Library size to the absolute minimum

1. If you are using Dropbox then you might want to take a look at your /Dropbox/Camera Uploads/ folder, since it might contain many images that may or may not be in your iPhoto folder. (Make sure you have checked the option in iPhoto -> Preferences -> Advanced -> Importing: Copy items to the iPhoto Library). Then import them all into iPhoto (we will remove all duplicates later). Then delete them from /Dropbox/Camera Uploads/ . Then go to Dropbox preferences and uncheck “Enable camera uploads for: ..”, so you won’t have excess images clogging up your Dropbox again. If you still want a cloud backup of your iPhoto folder, you can always just put your iPhoto library into Dropbox or Google Drive. 2. Use PhotoSweeper to eliminate all duplicate photos. Set it to find duplicates by compiling and comparing bitmaps with 48x48 size. That way you are sure to find images with identical content and not just images with same names, dates and other metadata. 3. In iPhoto: Select all photos. File -> Export -> JPEG Quality: Low (smallest size) + Check include title and keywords and location information. + Size: Full Size (this is the resolution of the image, not the size of the file). With JPEG Quality: Low (smallest size), you still likely won’t be able to tell the difference from the original, when viewed on a computer screen. For storing pictures for future display on a computer, a small digital photo frame, or on Facebook, you don’t need a larger size. You will likely not see the difference with the naked eye, and Facebook has limits on image size you can upload anyways. You only need to export a larger image size if you intend to print out the images or make a full size poster (A4, A3 format) of them. Then you would see the difference. For your best images, which you might want to do that with, you would want to export them with a higher quality, so you don’t loose the quality that you need. 4. Delete all images in iPhoto (if they are in an album/folder in iPhoto, then go to it, click CMD+A to select all the photos there, click CMD+ALT+BACKSPACE to move them all to the trash. NB: Just using the Backspace/Delete key or CMD+Backspace or right clicking and selecting "Move to trash" will actually not move them to the trash! It will only remove them from the album.) After deleting the images from iPhoto then import the compressed images from the place you exported them to.

5. Find the iPhoto Library file -> Right click -> Show package contents -> Thumbnails -> Search for “_1024″ (match the file name). Then select all and move them to your system trash. These are large thumbnail images iPhoto generates (when you import photos) to be able to show a 1024 pixel version of the image quite rapidly within iPhoto even for large (>1 MB) files. When you have compressed your image files to JPEG Quality: Low, they should only be 200-600 kb in size and so the 1024 pixel thumbnail won’t be any smaller or faster, so those thumbnails are not needed any more. 6. You might also want to run CleanMyMac to delete your master image files, if you are confident that your edits to your images can be permanent. The master images are just kept by iPhoto as a duplicate backup in case you want to revert edits to any images. Keeping them will easily double or triple your library size, so you might want to delete them. Finally: Make sure you have emptied the trash in iPhoto and after that also emptied the trash on your system.

[magnemg]

How to solve Chrome not remembering sign-in, and mysteriously closing and reopening itself on startup.

The problems:

- You suddenly always have to log in to Google’s services (Gmail, Calendar etc.) every time you open Chrome, and you have to log into your user profile in the upper right corner of the Chrome window every time you open chrome as well (it may also show a yellow warning sign before you log in). Chrome doesn’t seem to remember your logged in session or your logged in user, no matter if you check “remember me on this computer”, and no matter what you do with the settings. - Chrome says “Account sign-in details are out of date.” under the “Sign In” headline in chrome://settings/ . It seems like Chrome or your user profile is perpetually out of sync, or that your browser user profile is corrupted. - Chrome mysteriously automatically closes and reopens only seconds after you open it. My system, at the time:

Mac OSX 10.9 Chrome version 41.0.2272.118 Proposed solutions I tried, without success:

- Disconnect my Google Account. - Delete my user profile under the “People” headline in chrome://settings/ - Completely delete Chrome and all user data, and re-download and re-install Chrome. The real solution, in my case:

I found out that I somehow had gotten some malware/spyware installed called PremierOpinion, that had added itself as a green square in my OSX task bar, and was running as a background process which I could see with “Activity Monitor”. It had also installed itself to the Applications folder. PremierOpinion calls itself “researchware”, but it hijacks the Chrome browser, which is supposedly what happens when Chrome closes and reopens when you start it. It supposedly logs your internet use, so you should probably change the passwords you typed in while experiencing the aforementioned problems.

After deleting it from the Applications folder (you should use CleanMyMac or some other software to ensure you actually delete all files it may have placed in your system/user folders), and using “Force Quit” to remove the background process, Chrome would start up as normally (open only once) and I could connect my Google Account and sign in, and it would remember me the next time I opened my browser. I also changed the passwords that I used while experiencing the problem (for Google, as well as any other services), as they have to be considered compromised.

If this was helpful to you, please tweet the link to this post, so it shows up high in Google results and that others may find it as well.

[magnemg]

How to extract only the necessary /usr/bin files from the Mavericks OSX installer

Download Mavericks OSX (version 10.9) from the App Store (requires >= 5.35 GB of HDD space). You can actually skip running the entire "Install OS X Mavericks.app" installer after it has downloaded. Instead, you can use Pacifist to open the installer package, and extract only the /usr/bin files you need.

open /Applications/Install\ OS\ X\ Mavericks.app/Contents/SharedSupport/InstallESD.dmg open -a Pacifist /Volumes/OS\ X\ Install\ ESD/Packages/Essentials.pkg

Then, in Pacifist, navigate into 

Contents of Essentials.pkg/usr/bin/

and find the original system files there. Extract them to a folder on your computer, like this folder, for example:

mkdir -p ~/Downloads/Restoring original OSX Mavericks CLI shims in usr bin folder/2. Overwrite usr bin with these OSX Mavericks CLI shims extracted from 'Install OS X Mavericks.app'/

If you need to use these files to overwrite the files in your /usr/bin folder, then follow the instructions at the bottom of this blog post. NB: You should generally never overwrite or delete files within your /usr/bin folder.

[magnemg]

How to completely uninstall osx-gcc-installer without reinstalling OSX

You may have installed osx-gcc-installer in an attempt to solve another problem you may have had. I did so, because of problems using capybara-webkit with MP4 files on OSX Mavericks. Realizing that using osx-gcc-installer didn't fix your problem, and that installing it was probably a mistake, because you are unable to uninstall it, and no matter what you do, "xcode-select --version" won't show and updated version (newer than 2003), since osx-gcc-installer broke the system references that OSX uses to link to updated xcode-select versions, then you will find this guide helpful.

osx-gcc-installer basically overwrites a lot of files in your /usr/bin folder (which you should never do, unless you want to potentially break a lot of stuff, and have to reinstall your OS).

 Ideally,

you should be able to uninstall osx-gcc-installer (

https://github.com/kennethreitz/osx-gcc-installer

) by installing Xcode on top of it (like the author has suggested), or running the separate Apple command line tools (CLT) installer, but that doesn't work, for the reasons mentioned in this article.

In my case, Homebrew still gave:

Warning: You seem to have osx-gcc-installer installed. Homebrew doesn't support osx-gcc-installer. It causes many builds to fail and is an unlicensed distribution of really old Xcode files. Please run `xcode-select --install` to install the CLT.

The problem was that 'xcode-select --install' wouldn't show the installer, because osx-gcc-installer had overwritten /usr/bin/xcode-select with an old version that didn't have that command. Even ignoring the install command, and downloading the command line tools from Apple, and installing them, didn't remove the Homebrew warning.

Note that the Homebrew warning is important and shouldn't be ignored, as osx-gcc-installer will break many builds, for example building Qt 5 (with 'brew install qt5'), as I encountered, when I had to do that, to solve this other problem: http://magnemg.tumblr.com/post/113251336220/how-to-solve-a-capybara-webkit-and-video-js

osx-gcc-installer will overwrite files in you /usr/bin folder, which are system files. A good rule of thumb is to never overwrite files in the /usr/bin folder, because those are system files, and doing that may cause unexpected errors and force you to reinstall OSX. 

You will then run into problems with install_name_tool, as people have reported here:

https://github.com/Homebrew/homebrew/issues/29989

And Homebrew will tell you that:

Warning: You have an outdated version of /usr/bin/install_name_tool installed.

So, how do you make a complete uninstall of osx-gcc-installer, with no remains, then?

See this blog post, and see my comments below here before you do anything:

http://tumblr.martinemde.com/post/56563793964/remove-osx-gcc-installer-and-uninstall-xcode

It suggests backing up your system before running the uninstall script. But if you don't have Time Machine, or the HDD space to use it, a simpler way of backing up your /usr/bin files is simply to copy them into a bakup folder:

sudo mkdir /usr/bin.bak sudo cp -R /usr/bin/ /usr/bin.bak/

Run the uninstall script mentioned in that blog post (also downloadable here [1]). If the script doesn't run, or tell you that "Command Line Tools not installed.", as it did in my case, you can manually execute the remove commands like this:

sudo mv /usr/bin/clang ~/.Trash

And so on, for all the files you find in the listing at the bottom of that blog post. I didn't find all of the files listed there, as I must have uninstalled some of them previously. No worries.

The osx-gcc-installer overwrites clang and that is at least one of the files (if not the most important file, beside install_name_tool) that will give the mentioned Homebrew warning ("You seem to have osx-gcc-installer installed"). The script merely deletes a lot of the files in your /usr/bin folder that osx-gcc-installer overwrote, but your system depends upon them, so you need to restore them somehow. And reinstalling Xcode or the standalone CLT after running the script won't restore all those files, so OSX and Homebrew still won't be able to find the CLT developer tools in the /usr/bin folder.

The files osx-gcc-installer has overwritten, and that you have now deleted with the script, are actually shims, which means they are references to the files that OSX uses and actually updates on system updates. So when overwriting those, you are stuck on that old version forever (‘xcode-select —version’ gave version 2003 in my case, when I had installed osx-gcc-installer, and even after I had ran the uninstall script).

Reinstalling the OS is the recommended solution in this case (sadly). See: https://github.com/Homebrew/homebrew/issues/29989

But I presume you don't want to download the entire >= 5.35 GB of HDD space which Mavericks OSX from the App Store requires. (If you actually do want to do that, or the file "/Applications/Install OS X Mavericks.app" already exists on your computer, then you can follow these instructions instead.)

Instead:

Download only the necessary OSX Mavericks usr/bin files here

What if you don't have enough HDD space to download the entire Mavericks OS again, from the App Store?

I have created a torrent file, which contains the files from the official Mavericks OSX download. The OSX files were extracted it by following this procedure.

So download the following torrent file to your ~/Downloads folder, and unzip it.

Torrent file: https://archive.org/details/completely_uninstall_osx-gcc-installer

(Please keep seeding the torrent, so others can download it in the future as well. It only takes a few KB so it shouldn't even be noticable on your bandwidth.)

Then copy them into your /usr/bin folder, using the command below:  (But first ensure that you have taken a backup of your usr/bin folder, like suggested above.)

sudo cp -R ~/"Downloads/Completely uninstall osx-gcc-installer/Restoring original OSX Mavericks CLI shims in usr bin folder/2. Overwrite usr bin with these OSX Mavericks CLI shims extracted from 'Install OS X Mavericks.app'/" /usr/bin/

NB: Notice the -R option here, which is important to correctly copy the aliases, like c++ which is 7 bytes, and not 14 KB which it would be without that option.

You're done! You've removed all remains of osx-gcc-installer, and replaced them with the original OSX files, so your system should be stable again, and able to incorporate updates to the CLT either through Xcode or the official standalone CLT. Check that your 'xcode-select --version' now reports version 2333 or higher. Running 'brew doctor' again and it should give no warnings, or just a warning about an outdated Xcode, which can safely be ignored.

[1] - The script from martinmede's blog post is also included in the torrent.

Run it with this command:

sudo sh ~/"Downloads/Completely uninstall osx-gcc-installer/Restoring original OSX Mavericks CLI shims in usr bin folder/1. Uninstall osx-gcc-installer with this script/remove_CLI_tools.sh"

[magnemg]

How to solve a capybara-webkit and video.js problem with MP4 playback on OSX 10.9.5 Mavericks

I was trying to get my rspec tests to play an MP4 video, and fire a tracking event when the video had completed. The tests would run, but time out or don't find the video on the page which my rspec tests were trying to test. The problem was that Webkit wouldn't run the video, because it couldn't find the video element (or the play button). Basically the tests would time out (or try around 70 find calls, usucessfully) before finishing with an error:

    Capybara::ElementNotFound:       Unable to find css "div.vjs-big-play-button"

or

Capybara::ElementNotFound:       Unable to find xpath "//div[contains(@class,\"vjs-big-play-button\")]"

If you're interested, you can see my test code and full error output here:

https://gist.github.com/redbar0n/abb2524c9a883b11b44a

The line in my test apparently causing the error was:

page.find("div.vjs-big-play-button").click #click play video button

It would fail even before the .click was attempted executed, so the problem was in the find call. The error also said could not find element or something like that.

The tests would run fine on my partner's machine, which was running Ubuntu. So the problem had to be in some OS dependencies that Webkit has on OSX.

Running with selenium instead of capybara-webkit also worked fine (though took a lot longer to run, so it's unsuitable when you want to run the tests often, like in a CI environment). With :selenium you have to watch the tests run through in the browser, but I just wanted the tests running in the background like :webkit does. For future reference, here is my system specs at the time:

Video file I was trying to play: MP4 Mac OSX 10.9.5 Firefox 36.0 Ruby 2.0.0 Rails 3.2.19 XQuartz 2.7.7 Flash 16.0.0.305 (Shockwave Flash 16.0 r0) capybara (2.4.4)     mime-types (>= 1.16)     nokogiri (>= 1.3.3)    rack (>= 1.0.0)    rack-test (>= 0.5.4)    xpath (~> 2.0) capybara-webkit (1.3.1)    capybara (>= 2.0.2, < 2.5.0)    json

I tried running spec_helper.rb with this setting, to get some proper error message:

Capybara.javascript_driver = :webkit_debug #:webkit #:selenium

But it merely gave me:

http://127.0.0.1:54892/assets/projectname/unit.js|11214|VIDEOJS:   undefined|0|TypeError: 'undefined' is not an object

So I tried replacing video-js.js and the other videojs files with the latest version, but with no change in outcome. Also tried adding sleep commands before and after the `page.find("div.vjs-big-play-button").click` line in the tests, but nothing changed (except that I confirmed that it would sleep (wait) before and after it reports the  `undefined|0|TypeError: 'undefined' is not an object` problem). I tried circumventing the problem by using the poltergeist gem instead of capybara-webkit, since it uses PhantomJS and not Webkit as the JS engine in the tests. But I soon discovered that PhantomJS doesn't support video/audio, because they want it to be a small headless and independent test suite, and don't want to ship the video/audio codecs with it. So that didn't work.

Tried downgrading capybara-webkit, without success. I had to go through a bit of dependency hell to downgrade to capybara-webkit 1.1.1 which someone suggested might be a solution. But it wasn't a solution, after all. So don't try that. If you really want to downgrade still, for any reason, here are the versions that I discovered work with each other (for what it's worth):

gem 'capybara-webkit', '1.1.1' gem 'capybara', '2.1.0' gem "rspec-rails", '2.14.1'

But you don't need to do this downgrading, as said. I figured out that the problem lies in the fact that capybara-webkit uses (and recommends) Qt 4.8.6, which was in fact installed on my machine. capybara-webkit depends on a WebKit implementation from Qt. So I upgraded to Qt 5.4.1 by running:

brew uninstall qt brew install qt5

But I realized that Qt 5.4.1, as opposed to earlier versions, requires the full Xcode to be installed, to be able to build itself. (It will probably remain a requirement for future Qt versions.) This presents itself in some annoying error:

You may need to confirm the license agreement by running /usr/bin/xcodebuild.

But since I didn't have space on my hard drive to download and install the > 5 GB install that Xcode 6.1.1 was, I tried circumventing it, by just downloading the parts of Xcode (the files like xcodebuild etc.) that qmake presumably would need to build Qt 5. I found osx-gcc-installer, which promised to be an extract of only the relevant files. So I installed that. That was really stupid. It basically overwrites a lot of files in your /usr/bin folder (which you should never do, unless you want to potentially break a lot of stuff, and have to reinstall your OS). The files it overwrites are actually shims, which means they are references to the files that OSX uses and actually updates on system updates. So when overwriting those, you are stuch on that old version forever ('xcode-select --version' gave version 2003 in my case, when I had installed osx-gcc-installer).

Don't install osx-gcc-installer! But if you already have, and need to uninstall it: See my blog post on how to uninstall it completely without having to download and reinstall OSX entirely. After successfully uninstalling the osx-gcc-installer, and restoring the system files it overwrote, to their original state, I had to "bite the sour Apple" ('bite the bullet' in english, but the norwegian version of this idiom fits remarkably well here..) and find a way to download the full Xcode (there is no way around it, believe me, I have researched it thoroughly).

After installing Xcode 6.1.1 from the App Store, I then had to redirect xcode-select to use the Xcode version of the command line tools (CLT), since it is now again bundled within Xcode itself. I had the separate CLT installed from before, so that gave me:

xcode-select --print-path /Library/Developer/CommandLineTools

I had to refer to the Xcode version of the command line tools, like this:

sudo xcode-select --switch /Applications/Xcode.app/Contents/Developer/

And check that it worked, by running 'xcode-select --print-path' again. Then I deleted the /Library/Developer folder and its contents, since it only had the CommandLineTools there, which was the old separate command line tools install, and I don't want that if I already have it in the full Xcode install.

Then I was able to install Qt 5:

brew install qt5 brew doctor

And I linked the qt5 libraries Homebrew suggested in the post install message.

Then I tried running:

gem install capybara webkit

But it reported:

sh: qmake: command not found [...] Command 'qmake -spec macx-g++ LIBS\ \+\=\ -L/usr/local/opt/libyaml/lib\ -L/usr/local/opt/readline/lib\ -L/usr/local/opt/libksba/lib\ -L/usr/local/opt/openssl/lib' failed extconf failed, exit code 1

So I ran:

ln -s /usr/local/Cellar/qt5/5.4.1/bin/qmake /usr/local/bin/qmake

To create a symlink so that qmake -v would run and give me:

QMake version 3.0 Using Qt version 5.4.1 in /usr/local/Cellar/qt5/5.4.1/lib

To be on the safe side, I ensured to uninstall all potential old capybara-webkit gems I had installed:

gem uninstall capybara-webkit

When running 'gem install capybara-webkit' again, it would all work. Yay!

› gem install capybara-webkit Building native extensions.  This could take a while... Successfully installed capybara-webkit-1.4.1 1 gem installed

Then I was able to run 'bundle install' successfully, without further errors.

So, with the new Qt version installed, which uses an updated Webkit implementation, then capybara-webkit would work, and my test would finally be green! Conclusion: capybara-webkit does work with video on OSX Mavericks, but not with Qt 4.8.6, only with Qt 5. Moral: Don't use osx-gcc-installer, or ever overwrite files in /usr/bin I hope this helps anyone else, as this was a nasty problem that took me a good while to finally figure out how to solve.

[magnemg]

How to setup your DNS provider's root (apex) domain to route to Heroku

For Norwegians, this also answers the question: "Hvordan bruke Domeneshop med Heroku" Problem: 

Difficulty setting up your DNS to route http://yourwebsite.com to http://yourapp.herokuapp.com You cannot route http://yourwebsite.com to an IP address, because with a dynamic app hosting service like Heroku, that can change at any time. Cause: 

Your DNS provider doesn't support using CNAME routing (because it is against the DNS specification), and doesn't support the modern ALIAS or ANAME records. This is the issue with the DNS provider Domeneshop.no, for instance.

Solution:

If it exists, then delete the A- or AAA-record for http://yourwebsite that points to an IP address.

Set up WWW-forwarding from http://yourwebsite.com to http://www.yourwebsite.com.

Ensure that you have a CNAME record set up for http://www.yourwebsite.com that routes it to http://yourappname.herokuapp.com.

Now, when people go yo http://yourwebsite.com they will be routed to http://www.yourwebsite.com/ which in turn will take them to http://yourappname.herokuapp.com (but the http://www.yourwebsite.com will still be showing in their address bar). This is also suggested by the Heroku documentation:

https://devcenter.heroku.com/articles/custom-domains#subdomain-redirection (which also talks about the potential problems with HTTPS / SSL using this method)

[magnemg]

How to use localeapp.com with Heroku

Localisation (i18n)

All  app translations are stored in <locale>.yml files, and synced to localeapp.com. Only those files are synced. Examples: en.yml nb.yml

System translations (for rails, rails_admin, or language defaults like date and time localisations) should be stored in their own .yml files and thus NOT synced to localeapp.com. This will keep your translation key count down.

Heroku will sync (pull) automatically from localeapp.com every 60 seconds, or on page refresh. If do this implementation: this: https://github.com/Locale/localeapp/wiki/Deploy-a-Locale-Backed-Application-to-Heroku

The daemon is mainly made for for privately managed servers, where you want to pull translations regularly. But it can also be used on localhost. To pull all up-to-date translations, run (works only as a pull): `localeapp daemon`

The localhost app will sync on page refresh, for that specific page. So the daemon is not strictly needed, unless you have someone besides yourself constantly updating translations on localeapp.com. Pushing local changes to localeapp.com: If you change the en.yml/nb.yml files locally you must remember to run: `localeapp push config/locales/<locale>.yml` for the respective locale files you updated (replace <locale> with either en or nb or similar). 

That is the only way localeapp.com will get your local changes to those files.

After pushing, check for potential errors here: `http://www.localeapp.com/projects/<localeapp_project_id>/imports` (localeapp_project_id is the project ID on localeapp.com, you can see it from the URL when you're logged in there and viewing the translations for your project.) since the localeapp push won't show errors in the terminal directly.

NB - Gotcha: Beware that after updating <locale>.yml in a text editor, you have to successfully push to localeapp.com, before refreshing any view in the browser. Because refreshing a view in the browser will fetch the translations from localeapp.com, and overwrite the <locale>.yml file, thus effectively undoing your changes.

[magnemg]

Should you go straight from paper sketches to HTML/CSS, or via a design tool like Sketch, Adobe XD or Figma first?

The post has been moved to my dev.to blog: https://dev.to/this-is-learning/should-you-go-straight-from-paper-sketches-to-htmlcss-or-via-a-design-tool-like-sketch-adobe-xd-or-figma-first-2ho9

[magnemg]

How to deal with an overwhelming amount of tasks (in GTD and in life)

Posted on the GTD Facebook page:

"I have written down every work and personal task I need to do, including converting emails to action items and now I have 580 work tasks, 346 personal tasks, 266 tasks for my assistant and 117 honey-dos for my husband! I have them organized by project and date, but am feeling overwhelmed by the sheer volume of it all! Any advice? Thanks so much for your work. Answer from David Allen: Well, you have as many commitments as you have, and unless you want eternal subliminal stress, you need to get them objectively out of your head and reviewable. As you’ve discovered, your next task to get more stress-free is to determine which ones are really “someday” vs. which ones need to be on the front shelf. Essentially, everything that you’re not doing at any moment is “someday,” but the psyche feels much better when you have made some distinctions between the active ones that you really want/need to get done within a reasonable time vs. those that can wait. Ultimately you’ll have to decide what kind of overview/map you need and want to see, to feel OK about what you’re doing. So there’s no right or wrong answer about any of this—only what’s most workable for you.

I've had this exact same problem, and the solution isn't accepting that the weight of the world is on your shoulders.

Rather, the solution is: To shrug.

Bear with me..

The problem you're facing is:

Important tasks drown among unimportant ones. Same goes for urgent/not urgent tasks. You're not able to get a proper and manageable overview over what you need to do.  

"Long lists don't get done." It's a guilt trip. You feel bad about yourself when you're not able to make the list become noticeably smaller.   

GTD doesn't really give any good advice on "what tasks NOT to write down". The only mention in GTD I've found is: ".. if you're like most people, having some type of self-regulating mechanism will help you become more aware of what you want to keep and what you should just get rid of" [regarding the read/review list] (p148). What if you don't have that self-regulating mechanism? How do you obtain it, before the system overflows?

The underlying cause is probably:

You're a creative person. This also plays out in coming up with stuff that you could potentially do. (GTD actually encourages you to be creative when writing your lists, often giving you simply too much noise.)  

You start with "Yes". And you're overly optimistic about your own capabilities, and how much time or effort tasks actually require. You're ambitious. You want to do more than you're able to get done (appetite > ability).  

You're afraid of missing out on something, so you write everything down, to not risk forgetting something. This is loss aversion at it's finest.

I've found the following principles work as a solution:  

Forgetting and "missing out" can be a blessing.

Forgetting is actually the best way "to get something completely off your mind" (which you might recognize is the stated purpose behind GTD). Writing stuff on a list will only get it off your mind until the next time you look at that list. If you look at the list often, then that task will consume your attention more often (and give you a guilt-trip too). Attention is your scarcest resource. Before you wrote everything down, you were perfectly able to forget all the less important tasks, or someday/maybes, and that didn't cause you any pain, did it? Being able to forget stuff makes you able to give your full attention on the stuff that is really important. 

The world won't end if you miss out on doing certain tasks every now and then. Don't succumb to the worry and Fear Of Missing Out. Forgetting the mundane, will help you taste life more (and not living it through your lists). Accept the fact that you will be missing out possible-things-to-do all the time; on the web, in relationships, and in life in general. No matter what you do, or how much you do, there's always an abundance of things you could potentially also do, that you won't be able to. It's actually a good thing that you can't do everything, because that forces you to choose what you value. That will set you apart. Go on a todo-diet. See how many things you could get away with not doing, and experience the fact that life goes on, pretty much as normal, even still. People will ask you a second time if you've forgotten their request, and they won't be mad if you tell them they have to wait. It's totally OK to not "jump onto" a email/request/task the minute you see it, or to be constantly "hovering" over your inbox. People really don't expect you to do it immediately (in case they did, they would/should have texted or called you). And the next time life hands you that "thing you should do something about", or you experience "that problem you should fix" life will gently remind you, automatically.

"Start with No" / "Don't write it down".

Make actions work hard to get on your next action list. Make each action prove itself and show that it's a survivor. Make the initial response "not now" instead of "yes, lets write it down". (thank you 37signals). This is opposed to David Allens advice of writing everything down: "you need to get them objectively out of your head and reviewable". If the action keeps coming back (in your mind, or from others), then it's time to consider it. If you want to do it (in the light of all the other next actions you have, and while realizing that it will come at the expense of the others), then write it down. If someone requests you for something they're motivated to get from you, ask them to email you about it, instead of writing it down yourself at first. If they want it, it's their responsibility to go get it. If they forget (or don't bother) to to email you about it (or ask you again for it), then it's probably not that important to them. Viola, you just relieved yourself of remembering / tracking it. Being able to forget it made you able to focus on the stuff that was really important.

You just won't be able to do it all!

I find this to be unsound advice from David Allen: "you have as many commitments as you have".

Most of the stuff you deem you "need to do", you really don't. "Pleasing everyone, all of the time, is a surefire way to failure" someone once said. I'd extend this to pleasing your own apetite / ambition / wants all of the time. When your GTD-system inevitably crashes, or you realize that tasks have been lying on your next action list for weeks/months, you realize that it's really not a task you "need to do". Trash it, or put it into "someday/maybe" (then trash it from that list again if it's been there for several months).  

Break stuff into smaller lists.

I go against GTD's advice, and use a daily todo-list. I try to filter my next-action list in the morning, and make a smaller list of stuff I decide to do "Today". The tasks on that list are grouped by context, and with a lot of whitespace between the contexts (so the contexts look like separate sub-lists). There, a much more manageable list. If tasks carry over to the next day, I just use the same "Today" list as the day before (I just update it during my daily review of the next actions list).  The reason for this today-list is in having decided on an agenda for the day, and to:  Avoid drowning your important/urgent tasks in between the rest. Such a today-list is against the advice of GTD, but is actually in line with what David Allen stated above:

"[it] feels much better when you have made some distinctions between the active ones that you really want/need to get done within a reasonable time vs. those that can wait."

The key is realizing: You won't be able to glance at a list of 20+ tasks and quickly know what to do next!  

Trust your intuition more than your lists.

Your brain is the best filter in the world. I've often looked for "the answer" somewhere in my lists. It's funny how we look for someone or something (even our own lists/systems) to give us the answer. Instead of exerting the (painful, strenuous) effort of thinking ourselves.  When I realize I'm doing this, it's probably because I'm really low on energy, and default to less cognitive demanding tasks (like moving stuff up and down in my lists). Take a good break, and get something to eat. Sadly, but truly: There is no substitute for thinking yourself. Usually, "the answer" to what you should be doing right now, you somehow already know. It's usually what makes you feel accomplished / not accomplished at the end of the day. Use your lists only as a check, not as a "blueprint answer". Don't work "in" the list (sorting, prioritizing etc.). Let your intuition give you the answer instead. What I'm trying to do these days:

1. Look over the calendar and the list, for a rough overview. 2. Put the list away, take a break and contemplate what I really want to do, and how I should start it. 3. Without looking back on the list: Start working on a task that would make me feel accomplished (It's often an urgent and important task, or an important but not urgent task. Thank you 7habits). If it's hard to determine what the most important task is; just pick one. Better to get started than caught in analysis paralysis (yet again). 4. Later: Go back to the list, and check that I didn't forget anything. Let the list only serve as a check that you didn't forget anything important.  

Working on GTD or actually getting things done?

If you're spending a lot of time on managing your system/tasks, and not on getting things done, then you're doing it wrong. Set aside some time to reflect on how you could reduce your system to a minimum. System overhead is wasteful expenditure of attention. An optimal system, should get out of your way and get you going in the overall right direction, as fast as possible. You can course-correct underway.

So, when you're yet again overwhelmed with tasks, and it feels like the weight of the world is back on your shoulders: Shrug. Update: Here are some practical tips on what types of things to avoid keeping track of in GTD: 5 things to keep out of OmniFocus.