Debian Codegraph @

doc/INSTALL

raw
1 To install:
2
3 0. If you are reading this and suspect that you may not know what you
4 are doing, stop and check to see if an lcalc package is available
5 for your Linux distribution. The easiest way to install lcalc
6 "correctly" is to let someone else do it for you. Likewise if you
7 use Homebrew, Conda, or Nix to manage your software packages.
8
9 Otherwise, if you think that you want to configure, compile, test,
10 and install lcalc yourself... proceed.
11
12
13 1. Download and decompress the tarball.
14
15 If you're downloading version 2.0.0, for example, the tarball
16 should be called lcalc-2.0.0.tar.xz. It can be extracted on the
17 command line using tar:
18
19 $ tar -xJf lcalc-2.0.0.tar.gz
20
21 This will create a new directory name "lcalc-2.0.0" containing the
22 source code for the project. If, on the other hand, you are using a
23 graphical file manager (on, say, Windows or OSX), you can most
24 likely extract the tarball simply by clicking on it.
25
26
27 2) Configure the build.
28
29 Lcalc comes with a "configure" script that is used to... well, to
30 configure it. Run,
31
32 $ ./configure --help
33
34 in the project's directory to obtain a summary of its usage. The
35 design of every "configure" script is more or less the same. We
36 have included a documentation file called "INSTALL.automake" that
37 contains generic instructions for the script. If they're confusing,
38 you probably just want to use the default set of options by running
39
40 $ ./configure
41
42 with no arguments. Below we summarize the few lcalc-specific options
43 that the "configure" script takes:
44
45 * --with-pari
46
47 Adds support for elliptic-curve L-functions using the PARI library.
48 Requires that PARI be installed on your system in a location that
49 your compiler and linker know about.
50
51 * --enable-precision=<type>
52
53 Specify how much precision you want in your floating-point numbers.
54 The valid values for <type> are "double", "double-double",
55 "quad-double", and "multiple". The default is "double".
56
57 Both double-double and quad-double precisions require the David H.
58 Bailey's "qd" library <https://www.davidhbailey.com/dhbsoftware/>.
59 The "multiple" precision requires MPFR <https://www.mpfr.org/>.
60
61 * --enable-openmp
62
63 This experimental option compiles everything with support for
64 multithreading using OpenMP <https://www.openmp.org/>.
65
66 The "configure" script will check that your system has all of the
67 dependencies required to built lcalc. If any are missing, it will
68 inform you; otherwise, you're ready to build lcalc and libLfunction.
69
70
71 3. Compiling lcalc and libLfunction
72
73 This step is easy. Run,
74
75 $ make
76
77 in the top-level project directory.
78
79
80 4. Testing (optional)
81
82 Lcalc comes with a test suite for both the "lcalc" program and the
83 libLfunction library. To run the test suite, run
84
85 $ make check
86
87 in the top-level project directory after you've successfully built
88 the project (as in the previous step). Ideally you will see a
89 series of PASS results. If you see any FAIL results, please let
90 us know.
91
92
93 5) Installation
94
95 Now that the project is builtand you've (maybe) tested the result,
96 you're ready to install it. Again, this is easy if you've made the
97 right choices during the configuration step. Run,
98
99 $ make install
100
101 and everything should be installed to the locations determined by
102 the "configure" script. Note that, by default, this will try to
103 install files to e.g. /usr/local and will require root privileges.
104 If you want to install lcalc for only your user account, go back to
105 the configuration step and select a --prefix that is writable by
106 your user.
107
108 This will install the "lcalc" program to your chosen --bindir,
109 the libLfunction library to your --libdir, and its headers to
110 your --includedir.
111
112
113 6. Uninstallation
114
115 The installed project files can also be removed automatically, so
116 long as you keep the (unmodified) source tree around. At any later
117 time, you can run
118
119 $ make uninstall
120
121 to remove the files that were installed by "make install". If root
122 privileges were required during installation, they will also be
123 required for uninstallation.