The straight forward way to do such things is to first cd to the directory you want. Anyone who has ever worked with Java/Scala projects knows how deeply nested the sub directories could be, so naturally I was looking for a better solution. A while back, I stumbled upon autojump, which is kind of a Google’s /I’m feeling lucky/ feature for cd. It works great, as long as the top ranked entry is what you want. However, sometimes I wish I had more control over the result, specifically I wanted a better way to interactively filter the ranked list.

Thankfully, I found FZF, a superb command line fuzzy finder which helped me solve this problem (it is worthwhile installing FZF even if you don’t want to read the rest of this post. It is that good.). What I did was to simply combine FZF and FASD (FASD is a bit more flexible tool compared to autojump). The rest of this post assumes that you have both of them installed. In OSX, you can use brew to achieve this:

brew install fasd fzf

Once installed, ensure that the necessary init commands are placed in your shell rc. I use zsh, so I put them in my .zshrc. The next step is to create a command that will combine these two and act as a replacement for the cd command. For instance, if you use zsh, you can do:

unalias j 2>/dev/null
j() {
    [ $# -gt 0 ] && fasd_cd -d "$*" && return
    local dir
    dir="$(fasd -Rdl "$1" | fzf -1 -0 --no-sort +m)" && cd "${dir}" || return 1
}

What this function does is to define the shell command j. When you use j with an argument (e.g., j e), it simply passes the argument to fasd which will take you to the first result based on the keyword (in my case, the top hit for e is my .emacs.d directory). If I don’t use an argument to j, the function takes the ranked list from fasd and feeds it to fzf, letting you fuzzy search on the list!

Note that this is not limited to just the cd command. For example, I have defined command e that uses $EDITOR to open a file:

unalias e 2>/dev/null
e() {
    [ $# -gt 0 ] && fasd -f -e ${EDITOR} "$*" && return
    local file
    file="$(fasd -Rfl "$1" | fzf -1 -0 --no-sort +m)" && ${EDITOR} "${file}" || return 1
}

All these functions (and more) are defined in my .zshrc, which can be found here.

The key idea is really simple: make $HOME the git work-tree. The normal way of doing this would be to do a git init in your $HOME, but that would totally mess up git commands if you have other repositories in your $HOME (also, you probably don’t want your entire $HOME in a git repo). So, instead, we will create a dummy folder and initialize a bare repository (essentially a git repo with no working directory) in there. All git commands will be run with our dummy as the git directory, but $HOME as the work directory.

First Time Setup

Setting this method up the first time is really easy. First, let’s create our bare repository. I chose to name my placeholder .dotfiles (duh!)

mkdir $HOME/.dotfiles
git init --bare $HOME/.dotfiles

Now for the fun part. We will make an alias for running git commands in our .dotfiles repository. I’m calling my alias dotfiles:

alias dotfiles='/usr/local/bin/git --git-dir=$HOME/.dotfiles/ --work-tree=$HOME'

Add this alias to your .bashrc or .zshrc. From now on, any git operation you would like to do in the .dotfiles repository can be done by the dotfiles alias. The cool thing is that you can run dotfiles from anywhere.

Let’s add a remote, and also set status not to show untracked files:

dotfiles config --local status.showUntrackedFiles no
dotfiles remote add origin git@github.com:anandpiyer/.dotfiles.git

You’ll need to change the remote URL to your git repo. Now, you can easily add the config files you want to be in version control from where they are supposed to be, commit and push. For example, to add tmux config files, I’ll do:

cd $HOME
dotfiles add .tmux.conf
dotfiles commit -m "Add .tmux.conf"
dotfiles push

Setting Up a New Machine

To set up a new machine to use your version controlled config files, all you need to do is to clone the repository on your new machine telling git that it is a bare repository:

git clone --separate-git-dir=$HOME/.dotfiles https://github.com/anandpiyer/.dotfiles.git ~

However, some programs create default config files, so this might fail if git finds an existing config file in your $HOME. In that case, a simple solution is to clone to a temporary directory, and then delete it once you are done:

git clone --separate-git-dir=$HOME/.dotfiles https://github.com/anandpiyer/.dotfiles.git tmpdotfiles
rsync --recursive --verbose --exclude '.git' tmpdotfiles/ $HOME/
rm -r tmpdotfiles

There you go. No symlink mess.

My dotfiles are here for reference.

Step 1: Install FreePBX & Asterix

The first thing we are going to do is to wipe the old system and install FreePBX and Asterisk. To do so:

• Download the FreePBX distribution. I used version SNG7-PBX-64bit-1710-1 which bundles FreePBX 14 along with Asterisk 13, 14 or 15.
• Use UNetbootin or Etcher to create a bootable USB from the ISO image you downloaded and install it. During installation, you will be asked to choose the version of Asterisk. I chose the recommended combination (FreePBX 14 with Asterisk 13).
• You will need to create root password, remember this password.
• After the installation is completed, reboot the system and login using username root and the password you set earlier. You will then be presented with the web address of the FreePBX GUI.

Step 2: Configuring FreePBX

We now need to configure our FreePBX installation for use with Google Voice. In a different laptop/desktop, open the FreePBX GUI in a browser window. You will be asked to create an admin user the first time. Once logged in:

• It is recommended that you activate your FreePBX (free to do so). Once activated, you can access Admin -> System Admin where you will be able to do many things, including setting a static IP for your FreePBX installation.
• Go to Admin -> Updates and click Module Updates. Select Standard, Unsupported and Extended in repositories, and click on Check Online. The page will reload with your selection. Scroll down to Connectivity where you will be able to see Google Voice/Chan Motif. Click on it, select Download and Install. Scroll back up to the top of the page, and click on Process. The system will download and install the Google Voice motif. If an Apply Settings button appears at the top in red, click on it (do this every time this button appears).
• Go to Application -> Extensions and add a new PJSIP extension. Fill in name (e.g., 100), and a secret. Leave everything else to their default values and submit followed by apply settings.
• Go to Connectivity -> Google Voice and fill in your Google Voice username, password and number. Check Add Trunk, Add Outbound Routes and Send Unanswered to Google Voicemail. Once you submit and apply, check back the entry (the top right of the page will list your newly created entry). You should see the status as Connected. If you don’t, try enabling access to less secure apps in your Google account and retry.
• Go to Connectivity -> Inbound Routes. Add a new inbound route. In the DID field, put in your Google Voice number (10 digits only, e.g. 1234567890). In the Set Destination field, choose Extension and select the PJSIP extension we created earlier.
• At this point, you are ready to check if all this works. Install an SIP client (e.g., Zoiper SIP Softphone) on your phone, add your FreePBX IP as the SIP server. User will be the extension you created (e.g., 100), and the password will be the secret you set on this extension. Try making a call, it should go through Google Voice. If you call your Google Voice number, the SIP client should get the call. If both work, you are set.

Step 3: Configuring OBi100/110

Finally, let us point the OBi device to your FreePBX server.

• Login to OBITalk, and remove the (defunct) Google Voice entry.
• Recreate SP1 as a Generic Service Provider. Check Primary Line for Outgoing Calls. Use the FreePBX IP address as the proxy server, and 5060 as the port. Username and password would be from the extension you created in FreePBX (e.g. 100). Save it, and go to your dashboard. You should see Registered in the status.
• Now you should be able to make and receive calls via Google Voice through your OBi!

At a high-level, my literature management system is really simple. I keep a bibliography file to which I add papers that I find interesting. For each entry in this file, I link two things: the PDF of the paper, and notes on the paper. With this setting, I can quickly search for an entry, find the PDF and/or notes for the matching entry. All this is made possible by Emacs, with the help of a few packages. Let me explain this in a bit of details.

Basic Setup

The first thing to do is to set up a few basic things. As I mentioned, I have one bibliography file where I want new entries to go. I also need a folder to store all the PDFs and a way to store notes on each of these entries. For this task, I leverage the org-ref package. Here’s my org-ref setup:

(setq org-ref-bibliography-notes "~/org/ref/notes.org"
      org-ref-default-bibliography '("~/org/ref/master.bib")
      org-ref-pdf-directory "~/org/ref/pdfs/")

The code should be self-explanatory: bibliography entries are stored in a file named master.bib, while PDF files corresponding to the entries go in the pdfs folder. I chose to store my notes in a single org file notes.org. Note that it is entirely possible to have separate note files for every entry in your bibliography file. My org folder is synced with a cloud storage service.

Adding Entries

When I come across a paper that I’m interested in reading, I add it to master.bib. Most of the time, I can simply drag-and-drop the paper link or DOI to the bibliography file and let org-ref take care of creating the entry and downloading the PDF. Otherwise, I simply copy the bibtex entry and then run doi-utils-get-bibtex-entry-pdf to get the PDF.

Once I decide to read (and make notes on) a paper that is in my bibliography file, I then proceed to make an entry for it in my notes.org file. I do this in either of two ways: if I have the master.bib file open, I can simply go to the entry I want to read, and then use the org-ref-open-bibtex-notes command. As the name suggests, this command opens the corresponding notes entry in the org-ref-bibliography-notes file, creating a new entry if it can’t find one. If I don’t have the master.bib file open, I use the excellent helm-bibtex package to search through my bibliography files. When I find the entry I’m looking for, I open notes (or create it) with helm’s available actions on the selected entry. For this to work, we need to let helm-bibtex know where the notes and PDF files are:

(setq bibtex-completion-bibliography "~/org/ref/master.bib")
      bibtex-completion-library-path "~/org/ref/pdfs"
      bibtex-completion-notes-path "~/org/ref/notes.org") 

You can customize the fields in the entry created in notes.org by both org-ref and helm-bibtex to your liking.

Taking Notes

Now that I have an entry in my notes.org file for the paper that I want to read, the next step is to take notes as I read. You can simply type down your thoughts below the entry in notes.org, but I like seeing the notes side-by-side with the page on which it was made. For this purpose, I use the interleave package. To use the interleave package, all I need to do is to add the INTERLEAVE_PDF property to the PROPERTIES section of the entry in my notes.org file. Then, to open the PDF, I place the cursor on the title, and invoke interleave-mode command. This will display the PDF side-by-side with my notes, where I can navigate, see or add notes on any page.

Searching

The combination of org-ref, helm-bibtex and interleave gives me great flexibility. I can easily search for entries in my bibliography files using helm-bibtex. From there, I can open the paper and/or its notes. Since I use Emacs to write papers, org-ref makes it easy to insert citations, look up notes and download papers as I write. Having the notes in plain text lets me search through them easily. Finally, as a bonus, org mode lets me easily export my notes in several formats.

To summarize, having the literature management system tightly coupled with my paper writing environment (Emacs with AUCTeX) worked quite well for me.

• Track your cash flow. You should know where your money is going. I use Mint and Personal Capital for this purpose.

• Once you understand your spending patterns, create a budget. Have a goal in mind when creating the budget. For instance, my goal is to save a particular amount of money from every pay check. The good folks at YNAB is giving away their highly acclaimed software, You Need A Budget (YNAB), for free for students.

• It is hard, but make efforts to stick with your budget. Level app helps me with this, it shows me how much I can spend without going over my budget.

• Bank with an institution that does not ridicule you with miniscule rates and outrageous maintenance fee. I use Ally bank.

• If you manage to save money, invest your savings. With long-term inflation rate above 3%, a savings account is a bad investment choice. WiseBanyan lets you invest as low as $10 a month in a combination of stock and bond ETFs for free. Robinhood lets you trade stocks for free. The stock market has returned an average of over 7% historically. Assuming this holds in the future, investing $500 every month will make your net worth exceed the median American middle class by the time you graduate!

• Sign up for good cashback credit cards such as the Chase Freedom card. If you have more than one credit card, an interesting app to use is Wallaby, which tells you the best card to use when you shop.

• Cellphone service is a basic necessity, but gone are those days when you had to spend a fortune on a decent plan. For the frugal folks, FreedomPop offers FREE service.

• While it is convenient to order things from Amazon and have it delivered to your home, many everyday essentials such as toothpaste, mouthwash, soap, body wash etc can be obtained absolutely free if you are willing to do a little bit of couponing along with some help from store specific blogs such as IHeartCVS and IHeartWags. However, I must warn you that couponing is very additctive.

Turns out the issue occurs because of a modification during Windows installation. Once Bootcamp Assistant (BCA) created the BOOTCAMP partition and restarted the system, I noticed that there was an additional small-sized partition created (~600MB) just before the BOOTCAMP partition. Instead of ignoring it, I deleted and recreated the bootcamp partition from Windows setup hoping the wasted space would be reclaimed. That was a bad idea.

When you do this, it confuses the hybrid GPT/MBR partitioning scheme used by BCA. Essentially, when you remove/resize/add partitions via Windows setup, that information gets updated in the MBR record, but is not synced to the GPT record. Thus, Windows will see the OSX partition, but not vice-versa.

The easy way to fix this without going through the pain of removing and reinstalling Windows from scratch is to sync the MBR info back to the GPT record. Here’s how to do it:

First, we need detailed information on the MBR record—where the partition starts, where it ends etc. For this purpose, use the gptsync tool. Homebrew has the gptsync package, but if you try to use it, you’ll end up with the following error:

    Warning: Unknown GPT spec revision 0x00010000 

This is because of 64-bit incompatability. A fixed version of gptsync aimed at OSX, extracted from rEFIt is available here. Just compile it using Xcode, and you’re good to go.

Running gptsync on my machine resulted in the following:

    > ./gptsync /dev/disk0
    
    Current GPT partition table:
     #      Start LBA      End LBA  Type
     1             40       409639  EFI System (FAT)
     2         409640    250801735  Mac OS X HFS+
     3      250801736    252071271  Mac OS X Boot
     4      252071936    500117503  Basic Data

    Current MBR partition table:
     # A    Start LBA      End LBA  Type
     1              1       409639  ee  EFI Protective
     2         409640    250801735  af  Mac OS X HFS+
     3 *    250802176    251518975  07  NTFS/HPFS
     4      251518976    500115455  07  NTFS/HPFS   

In the scenario above, the solution was to modify the GPT table to match the MBR table. Essentially, delete partitions 3 and 4, and then recreate it as per MBR. For this purpose, I used the gdisk tool available here. Once installed, run the tool (usage instructions are similar to fdisk, tutorial is available on the tool’s website). I used d command to delete partitions 3 and 4, n command to recreate them with first sector and last sector noted from MBR and type 0700. Once done, check the changes using command p, and verify for errors using v. If there are no errors, confirm the changes by writing them with command w, reboot and enjoy!

Turns out that this is easy to fix. Here’s how to make it work.

First, unpack the IPA (unzip -q <IPA>). This would create a directory named Payload and a few other files. Delete the existing signature in the extracted package:

    rm -rf "Payload/*.app/_CodeSignature" "Payload/*.app/CodeResources"

For the next step, you need two things: a valid provisioning profile and a valid certificate. Both can be obtained via the iOS developer portal.

Replace the provisioning profile first

    cp <PROVISIONING_PROFILE> "Payload/*.app/embedded.mobileprovision"

Then resign using your certificate (the certificate must be present in your keychain)

    /usr/bin/codesign -f -s "<CERTIFICATE>" --resource-rules "Payload/*.app/ResourceRules.plist" "Payload/*.app"

Now you can recreate the IPA and then drag & drop it in the XCode organizer to install it on the device!

    zip -qr <IPA_resigned> Payload

Edit: If you are looking for user friendly options, check out the script by Daniel or the app by Brian.

    Error: GLib-GObject-CRITICAL **: gtype.c:2720: You forgot to call g_type_init() 

This is essentially an issue with the latest version of Pango package. There are two ways to fix this:

• Force homebrew to go back to a lower version of Pango; this error seems to appear in versions 1.32.6 or later.
• Patch the gnuplot installation’s Cairo terminal as suggested here.

Note: If you are not getting cairo terminals in your gnuplot installation, reinstall it with cairo option enabled: brew install gnuplot --cairo

• Tips for Conference Papers by Gurmeet Singh

  A nice set of tricks that saves your life at the last minute during deadlines.

• Latex Usage Notes by Eddie Kohler

  Solutions for common Latex problems, tips on proper formatting, etc.

• Table Generator

  Easy way to generate latex source code for tables, supports booktabs style too!

• Detexify

  Find latex symbols by simply drawing them. Huge time saver when you are working on a deadline.

Recently, I found a workaround for this problem. Here’s the trick to syncing any folder in your computer: softlink the folder to your Dropbox folder!

If you’re using a Mac (or Linux), just use the following command:

    ln -s /full-path-to-folder/file-to-sync ~/Dropbox/desired-folder-name

That’s it, from now on, all file(s) in the actual path would now sync with Dropbox.