Building from source

Building from source

Please also see Installation for information about pre-built executables.

Miller license

Two-clause BSD license

From release tarball using autoconfig

Miller allows you the option of using GNU autoconfigure to build portably.

Grateful acknowledgement: Miller’s GNU autoconfig work was done by the generous and expert efforts of Thomas Klausner.

  • Obtain mlr-i.j.k.tar.gz from, replacing i.j.k with the desired release, e.g. 2.2.1.
  • tar zxvf mlr-i.j.k.tar.gz
  • cd mlr-i.j.k
  • Install the following packages using your system’s package manager (apt-get, yum install, etc.): flex
  • Various configuration options of your choice, e.g.
    • ./configure
    • ./configure --prefix=/usr/local
    • ./configure --prefix=$HOME/pkgs
    • ./configure CC=clang
    • ./configure --disable-shared (to make a statically linked executable)
    • ./configure 'CFLAGS=-Wall -std=gnu99 -O3'
    • etc.
  • make creates the c/mlr executable
  • make check
  • make install copies the c/mlr executable to your prefix’s bin subdirectory.

From git clone using autoconfig

  • git clone
  • cd miller
  • Install the following packages using your system’s package manager (apt-get, yum install, etc.): automake autoconf libtool flex
  • Run autoreconf -fiv. (This is necessary when building from head as discussed in
  • Then continue from “Install the following … ” as above.

Without using autoconfig

GNU autoconfig is familiar to many users, and indeed plenty of folks won’t bother to use an open-source software package which doesn’t have autoconfig support. And this is for good reason: GNU autoconfig allows us to build software on a wide diversity of platforms. For this reason I’m happy that Miller supports autoconfig.

But, many others (myself included!) find autoconfig confusing: if it works without errors, great, but if not, the ./configure && make output can be exceedingly difficult to decipher. And this also can be a turn-off for using open-source software: if you can’t figure out the build errors, you may just keep walking. For this reason I’m happy that Miller allows you to build without autoconfig. (Of course, if you have any build errors, feel free to contact me at – or, better, open an issue with “New Issue” at


  • Obtain a release tarball or git clone.
  • cd into the c subdirectory.
  • Edit the INSTALLDIR in
  • To change the C compiler, edit the CC= lines in and dsls/
  • make -f creates the mlr executable and runs unit/regression tests (i.e. the equivalent of both make and make check using autoconfig).
  • make install copies the mlr executable to your install directory.

The is simple: little more than gcc *.c. Customzing is less automatic than autoconfig, but more transparent. I expect this makefile to work with few modifications on a large fraction of modern Linux/BSD-like systems: I’m aware of successful use with gcc and clang, on Ubuntu 12.04 LTS, SELinux, Darwin (MacOS Yosemite), and FreeBSD.


Disclaimer: I’m now relying exclusively on Appveyor for Windows builds; I haven’t built from source using MSYS in quite a while.

Miller has been built on Windows using MSYS2: You can install MSYS2 and build Miller from its source code within MSYS2, and then you can use the binary from outside MSYS2. You can also use a precompiled binary (see above).

You will first need to install MSYS2: Then, start an MSYS2 shell, e.g. (supposing you installed MSYS2 to C:\msys2\) run C:\msys2\mingw64.exe. Within the MSYS2 shell, you can run the following to install dependent packages:

pacman -Syu
pacman -Su
pacman -S base-devel
pacman -S msys2-devel
pacman -S mingw-w64-x86_64-toolchain
pacman -S mingw-w64-x86_64-pcre
pacman -S msys2-runtime

The list of dependent packages may be also found in appveyor.yml in the Miller base directory.

Then, simply run which is a thin wrapper around ./configure && make which accommodates certain Windows/MSYS2 idiosyncracies.

There is a unit-test false-negative issue involving the semantics of the mkstemp library routine but a make -k in the c subdirectory has been producing a mlr.exe for me.

Within MSYS2 you can run mlr: simply copy it from the c subdirectory to your desired location somewhere within your MSYS2 $PATH. To run mlr outside of MSYS2, just as with precompiled binaries as described above, you’ll need msys-2.0.dll. One way to do this is to augment your path:

C:\> set PATH=%PATH%;\msys64\mingw64\bin

Another way to do it is to copy the Miller executable and the DLL to the same directory:

C:\> mkdir \mbin
C:\> copy \msys64\mingw64\bin\msys-2.0.dll \mbin
C:\> copy \msys64\wherever\you\installed\miller\c\mlr.exe \mbin
C:\> set PATH=%PATH%;\mbin

In case of problems

If you have any build errors, feel free to contact me at – or, better, open an issue with “New Issue” at


Required external dependencies

These are necessary to produce the mlr executable.

  • gcc, clang, etc. (or presumably other compilers; please open an issue or send me a pull request if you have information for me about other 21st-century compilers)
  • The standard C library
  • flex
  • automake, autoconf, and libtool, if you build with autoconfig

Optional external dependencies

This documentation pageset is built using Sphinx. Please see ./ for details.

Internal dependencies

These are included within the Miller source tree and do not need to be separately installed (and in fact any separate installation will not be picked up in the Miller build):

Creating a new release: for developers

At present I’m the primary developer so this is just my checklist for making new releases.

In this example I am using version 3.4.0; of course that will change for subsequent revisions.

  • Update version found in mlr --version and man mlr:
    • Edit, c/mlrvers.h, miller.spec, and docs/ from 3.3.2-dev to 3.4.0.
    • Do a fresh autoreconf -fiv and commit the output. (Preferably on a Linux host, rather than MacOS, to reduce needless diffs in autogen build files.)
    • make -C c -f installhome && make -C man -f installhome && make -C docs -f html
    • The ordering is important: the first build creates mlr; the second runs mlr to create manpage.txt; the third includes manpage.txt into one of its outputs.
    • Commit and push.
  • Create the release tarball and SRPM:
    • On buildbox: ./configure && make distcheck
    • On buildbox: make SRPM as in
    • On all buildboxes: cd c and make -f mlr.static. Then copy mlr.static to ../mlr.{arch}. (This may require as prerequisite sudo yum install glibc-static or the like.)
    • For static binaries, please do ldd mlr.static and make sure it says not a dynamic executable.
    • Then mv mlr.static ../mlr.linux_x86_64
    • Pull back release tarball mlr-3.4.0.tar.gz and SRPM miller-3.4.0-1.el6.src.rpm from buildbox, and mlr.{arch} binaries from whatever buildboxes.
    • Download mlr.exe and msys-2.0.dll from
  • Create the Github release tag:
    • Don’t forget the v in v3.4.0
    • Write the release notes
    • Attach the release tarball, SRPM, and binaries. Double-check assets were successfully uploaded.
    • Publish the release
  • Check the release-specific docs:
  • Notify:
git remote add upstream # one-time setup only
git fetch upstream
git rebase upstream/master
git checkout -b miller-3.4.0
shasum -a 256 /path/to/mlr-3.4.0.tar.gz
edit Formula/miller.rb
# Test the URL from the line like
#   url ""
# in a browser for typos
# A '@BrewTestBot Test this please' comment within the homebrew-core pull request will restart the homebrew travis build
git add Formula/miller.rb
git commit -m 'miller 3.4.0'
git push -u origin miller-3.4.0
(submit the pull request)
  • Afterwork:
    • Edit and c/mlrvers.h to change version from 3.4.0 to 3.4.0-dev.
    • make -C c -f installhome && make -C doc -f all installhome
    • Commit and push.

Misc. development notes

I use terminal width 120 and tabwidth 4.