Split some utils, logrotate, improve posix.

Author: cirosantilli

ctags/README.md

@@ -150,3 +150,5 @@ Supports 55 languages
 
     exuberant-ctags a.c
     less tags
+
+Seems to be the main Ubuntu package that offers a `ctags` utility.

desktop/README.md

@@ -227,7 +227,7 @@ Keep examining the properties, and print if changes happen:
 
 Modify key maps.
 
-For example, to exchange exc and caps lock:
+For example, to exchange ESC and Caps Lock:
 
 	f=~/.Xmodmap
 	echo "! Swap caps lock and escape
@@ -246,6 +246,14 @@ To make this happen every time at startup TODO broken?:
 
 The `~/.xsession` file could also be used depending on system
 
+##xinitrc
+
+Sourced in Ubuntu 12.04, but not 14.04.
+
+##xsessionrc
+
+TODO vs `xinitrc`?
+
 ##xdotool
 
 Send clicks and manage window properties from sh.

filesystem.md

@@ -118,7 +118,7 @@ Use the same username on new systems, and mount the partition automatically with
 
 30GiB is a good size for each root partition. Leave everything else for the home partition.
 
-#Partition table
+##Partition table
 
 Contained in the MBR.
 

gpu.md

@@ -50,10 +50,20 @@ Xorg is getting old and the future Wayland is considering the shift towards swit
 
 Bumblebee seems to be the best bet to support NVIDIA Optimus as of 2013.
 
-#nvidia
+#NVIDIA
 
-##nvidia settings
+##NVIDIA version names
 
-To check that the installation is working, use `nvidia-settings`, which monitors the GPU, and will show if the GPU is not properly installed.
+Each hardware has two names:
+
+- model: precise name of a single piece of hardware
+- codename: marketing name that may group several models
+
+For example, from the `lspci` output, in `GF108M [NVS 5400M]`:
 
-##nvidia settings
+- `NVS 5400M` is the model name
+- `GF108M` is the codename
+
+##NVIDIA settings
+
+To check that the installation is working, use `nvidia-settings`, which monitors the GPU, and will show if the GPU is not properly installed.

logrotate.md

@@ -0,0 +1,37 @@
+A log file is a file that stores notifications, often automatically made by programs that run on the background like servers or the kernel.
+
+Log file for system wide utilities are often stored under `/var/log`.
+
+Log files contain information that can be useful to debug the system.
+
+Most applications simply append to the end of a single log file.
+
+This has two problems:
+
+- you can run out of disk space
+- opening each file takes a very long time
+
+It would be easy to solve those two problems if it were possible to write to the beginning of a file efficiently: applications could just do that and truncate the file to a size, keeping the most recent messages at the top. Unfortunately there is no efficient way of writing to the beginning of a file: you would have to move everything forward to make up space.
+
+Logrotate solves both of those problems with the premise that more recent logs are more important.
+
+For example, you can configure logrotate to do the following:
+
+-   if log the file `a.log` is larger than 100Kb, zip it into `a.log.1.gz`.
+
+    This will make up room for the application to create a new log file.
+
+-   if `a.log.1.gz` already exists, create `a.log.2.gz`, and so on, up to a configurable limit, like `a.log.4.gz`.
+
+    After that, the `.gz` would be removed.
+
+This is why it is called logrotate: at the same time, it transforms:
+
+- remove `a.log.4.gz`
+- rename `a.log.3.gz` into `a.log.4.gz`
+- ...
+- rename `a.log.1.gz` into `a.log.2.gz`
+- zip `a.log` into `a.log.1.gz`
+- remove `a.log`
+
+Most distributions run logrotate by default from cron.

make/README.md

@@ -1,48 +1,37 @@
 Make allows you to:
 
-- make command line interfaces of the type:
+-   make command line interfaces of the type:
 
         make <something> a=b c=d
 
-    *very* easily
+    *very* easily.
 
-- only build outputs when inputs have changed
+-   only build outputs when inputs have changed
 
-    this may save lots of time when building large projects
+    This may save lots of time when building large projects.
 
-Make is specified by posix and implemented by GNU with extensions
+Make is specified by POSIX and implemented by GNU with extensions
 It is the de facto standard make tool for POSIX systems.
 
-#downsides of make
+#Downsides of Make
 
 The main problems of make are:
 
-- low cross-platform portability
-- its Yet Another Scripting Language to learn (and not a very powerful one at that)
+- not available on Windows
+- its Yet Another Scripting Language to learn, and uses Bash inside of it
 
-#alternatives
+#Alternatives
 
-Because of the downsides of make, many other make systems have been devised. None has yet dominated on most applications, but important ones include:
+Because of the downsides of make, many other make systems have been devised.
+None has yet dominated on most applications, but important ones include:
 
-- cmake
+-   CMake
 
-    Generates programs that make the project, including projects for IDEs.
-
-    For example, it can generated:
-
-    - POSIX `Makefiles` on Linux,
-    - `cmd.exe` build scripts for Windows
-    - CodeBlocks project
-
-    all from the same input file.
-
-    Makefiles for cmake (called `CMakeLists.txt`) are written in Yet Another Scripting Language.
-
-- rake
+-   Rake.
 
     Similar to `make`, Ruby implemented with makefiles written in pure Ruby.
 
-- apache ant
+-   Apache Ant.
 
     Written in Java, mainly used for Java project.
 
@@ -52,20 +41,20 @@ Because of the downsides of make, many other make systems have been devised. Non
 
 - <http://stackoverflow.com/questions/12669367/should-i-name-makefile-or-makefile>
 
-`make` tries the following names, in order: GNUmakefile, makefile and Makefile
+make tries the following names, in order: `GNUmakefile`, `makefile` and `Makefile`.
 
 `Makefile` has the advantages:
 
-- appears at the top of ls by default (because ascii uppercase comes before ascii lowercase)
+- appears at the top of ls by default (because ASCII uppercase comes before ASCII lowercase)
 - is more conventional
 
 `makefile` has the advantages:
 
 - sticks to the other widespread convention of naming everything lowercase
 
-#command line options
+#Command line options
 
-- `-j5`: let make run in 5 threads.
+-   `-j5`: let make run in 5 threads.
 
     *Be warned*: this will make standard output of all threads mix up so the stdout will be meaningless.
 
@@ -76,10 +65,10 @@ Because of the downsides of make, many other make systems have been devised. Non
     `a`, `b`, and `c` are run in parallel, it it might be the case that
     `c` *must* run only after `a` and `b`.
 
-#sources
+#Sources
 
-- <http://www.jfranken.de/homepages/johannes/vortraege/make_inhalt.en.html>
+-   <http://www.jfranken.de/homepages/johannes/vortraege/make_inhalt.en.html>
 
-    good first tutorial to make
+    Good first tutorial to make.
 
-- <http://www-cip.physik.uni-bonn.de/pool/infos/make/advanced.html>
+-   <http://www-cip.physik.uni-bonn.de/pool/infos/make/advanced.html>

posix.md

@@ -1,10 +1,8 @@
-Cheat on POSIX.
-
 For a cheat on the POSIX C API see: <https://github.com/cirosantilli/cpp/tree/99ad8ab9aee7bf2d8c47c9e25f4631eddb556ccf/posix>
 
 POSIX command line utilities shall not be discussed here because it is better to group utilities together with other non-POSIX utilities which have similar functions, for example discussing both `kill` (POSIX) and `killall` (non-POSIX) side by side.
 
-POSIX is means: Portable Operating System Interface for uniX.
+POSIX means: `Portable Operating System Interface for uniX`.
 
 POSIX is also known as: Single Unix specification (SUS)
 
@@ -24,7 +22,7 @@ Currently, GNU/Linux and mac systems are largely POSIX compliant but not certifi
 
 The specification if free to view.
 
-Has several versions. The last at the time of writting was made in 2008
+Has several versions.
 
 POSIX issue 7: IEEE formal name: `IEEE Std 1003.1-2008` highly recommended link: <http://pubs.opengroup.org/onlinepubs/9699919799/>
 
@@ -62,7 +60,7 @@ Examples:
 - `ls`
 - `cat`
 - `mkdir`
-- `c99` and `fortr77`: compiler interfaces for C99 and Fortran77!
+- `c99` and `fortr77`: compiler interfaces for C99 and Fortran 77!
 
 and tons of others which most people never heard of
 
@@ -110,8 +108,21 @@ The following variables have fixed purposes in POSIX 7:
 - `PWD`
 - `TMPDIR`
 - `COLUMNS` and `LINES`: current width of terminal
-- `SHELL`: this is *not* the current shell! It contains the path of the default shell.
-- `TERM`: analogous to shell
+- `SHELL`:               this is *not* the current shell! It contains the path of the default shell.
+- `TERM`:                analogous to shell
+
+The following variables determine the locale applications should use:
+
+- `LC_CTYPE`:    Character classification and case conversion.
+- `LC_COLLATE`:  Collation order.
+- `LC_MONETARY`: Monetary formatting.
+- `LC_NUMERIC`:  Numeric, non-monetary formatting.
+- `LC_TIME`:     Date and time formats.
+- `LC_MESSAGES`: Formats of informative and diagnostic messages and interactive responses.
+
+They are defined at: <http://pubs.opengroup.org/onlinepubs/009695399/basedefs/xbd_chap07.html>
+POSIX also offers interfaces to modify those variables. The exact location where they are implemented
+is not specified. E.g., on Ubuntu 14.04, they are stored under `/etc/default/locale`.
 
 The following variables don't have fixed purposes, but POSIX says that they must be used with caution:
 
@@ -203,7 +214,9 @@ A few of the most interesting suggests for argument and utility name syntax:
     Good discussion about command line interface conventions: <http://stackoverflow.com/questions/10818443/short-long-options-with-option-argument-is-this-some-sort-of-convention>
 
 -   `--` marks the last of the options
+
 -   `-` means stdin/stdout when the utility expects a filename as argument
+
 -   `-W` is reserved for vendor options
 
 ##Exit status

sort.md

@@ -0,0 +1,52 @@
+Sort line wise.
+
+Uses External R-Way merge.
+This algorithm allows to sort files that are larger than RAM memory.
+
+Sort `f1`, `f2` together line wise:
+
+    sort f1 f2
+
+Useful options:
+
+-   `-r` : reverse order
+
+-   `-n` : numbers
+
+-   `-u` : `uniq`
+
+-   `-t:`: set field separator to `:`
+
+-   `-k5`: sort by the Nth field.
+
+    Example:
+
+        sort -k 2,2n -k 4,4hr
+
+    - sort field `2` (from `2` to `2`) numerically
+    - sort field `4` by human sizes and reverse
+
+    From field 2 to 2, numerically, then from 4 to 4, human and reverse
+
+-   `-R` : randomize: pseudo reverse of sorting.
+
+-   `-h` : sort by human readable file sizes like `1k`, `2M`, `3G`
+
+-   `-f` : ignore case
+
+-   `-fs`: ignore case and put always upper before lower
+
+-   `-b` : ignore leading blanks
+
+-   `-uf`: remove duplicates case insensitive
+
+-   `-m` : suppose `f1` and `f2` are already sorted, making sort faster
+
+-   `-c` : check if is sorted
+
+#GNU extensions
+
+-   `-V`: sort dot separated versions numbers:
+
+        [ "$(printf '10.0\n9.10\n9.9\n' | sort -V)" = "$(printf '9.9\n9.10\n10.0\n')" ] || exit 1
+

tee.md

@@ -0,0 +1,25 @@
+POSIX 7
+
+Echo stdin to multiple files ant to stdout.
+
+To stdout and file:
+
+    echo a | tee file
+
+To file and stderr:
+
+    echo a | tee file 1>&2
+
+Append to file:
+
+    echo a | tee –a file
+
+To multiple files:
+
+    echo a | tee f1 f2 f3
+
+To multiple processes:
+
+    echo a | tee >(seqn 2) tee >(seqn 2) | tr a b
+
+Process are run in parallel so the output order is variable.

tmux.md

@@ -87,7 +87,7 @@ Window shortcuts:
 
 Pane shortcuts:
 
-- `n`: and `p`: next and previous window
+- `n` and `p`: next and previous window
 - `0`: to `9`: switch to given window ID
 - `%`: split horizontally
 - `"`: split vertically:
@@ -96,7 +96,7 @@ Pane shortcuts:
 - `o`: cycle through panes in order
 - `{}`: current pane's position with next / previous one
 - `!`: remove pane from window and open it in a window of its own
-- `x`: kill the current pane
+- `x`: kill / close the current pane
 
 #Send keys
 

tr.md

@@ -0,0 +1,27 @@
+POSIX 7
+
+Character-wise text operations.
+
+Replace `a` by `A`, `b` by `B` and `c` by `C`:
+
+    [ `echo -n cab | tr abc ABC` = CAB ] || exit 1
+
+Ranges work as expected. Convert to uppercase:
+
+    [ `echo -n cab | tr a-z A-Z` = CAB ] || exit 1
+
+POSIX character classes are understood. Remove non-alphanumeric characters:
+
+    [ `echo -n 'ab_@' | tr -cd "[:alpha:]"` = ab ] || exit 1
+
+-   `-c`: complement and replace. Replace all non `abc` chars by `d`:
+
+        [ `echo -n dcba | tr -c abc 0` = 0cba ] || exit 1
+
+-   `-d`: delete `abc` characters:
+
+        [ `echo -n dcba | tr -d abc` = d ] || exit 1
+
+-   `s`: replace multiple consecutive `a` and `b` by a single character:
+
+        [ `echo -n aabbaac | tr -s ab` = abac ] || exit 1