-
-
Notifications
You must be signed in to change notification settings - Fork 819
Binaries
As of v2.1.16
(and pull request #192) node-sqlite3
now defaults to installing a pre-built binary of node_sqlite3.node
(with a statically linked libsqlite3
) when available.
The Node.js project is discussing the design of a system for binary deployments and a build farm for C++ modules like node-sqlite3
. But since this does not exist yet the approach described below for node-sqlite3
should be considered a stopgap solution. Once a better, official system is available we can replace this approach, but until then this should help many users of node-sqlite3
get running quickly.
To force building from source do:
npm install sqlite3 --build-from-source=sqlite3
or
npm install sqlite3 --build-from-source
To request an (additional) arch
be installed that is different from the value of process.arch
for your running node version you can pass --target_arch
. For example, to install a 32bit binary on the 64 bit system do:
npm install sqlite3 --target_arch=ia32
We have not yet built binaries for all possible variants. Here are the ones that are most requested and not available as of Sept 10, 2013
:
Users that find a binary is not available for their platform are encouraged to trigger a custom source compile that will package a binary for later distribution. To do this run:
npm install sqlite3 --stage
This will drop a tarball (.tar.gz
) and a shasum (sha1.txt
) in the stage/
folder within the installed module folder. These two files therefore will be available at $HOME/node_modules/sqlite3/stage/Release/
.
You can also run npm install --stage
from within a node-sqlite3
checkout from github.
To submit your built tar.gz
please email it to dane - at - mapbox.com. In the future we plan to automate this.
Currently available binaries are listed here.
Binaries are versioned based on the node-sqlite3
version, node version, platform, and architecture. Given this amounts many versions we have not yet created binaries for all possible combinations.
Specifically, versioning works like this:
${configuration folder}/node-sqlite3-${MAJOR}.${MINOR}.${ABI}-node-v${ABI}-${platform}-${arch}.tar.gz
- configuration folder - 'Release' or 'Debug'
- node-sqlite3
MAJOR
andMINOR
match the semver values for the given release. TheABI
is a special incrementing letter that marks new versions of the c++ code. - The node versioning uses the C++
ABI
number rather node's semver string. This value is available in javascriptprocess.versions.modules
as of>= v0.10.4 >= v0.11.7
and in C++ as theNODE_MODULE_VERSION
define much earlier. Currently thenode-sqlite3
build scripts access this value only viaprocess.versions.modules
so for versions beforev0.10.4
thev8
MAJOR.MINOR
is used as a proxy. -
platform
matches node'sprocess.platform
likelinux
,darwin
, andwin32
-
arch
matches node'sprocess.arch
likex64
oria32
Below is a listing of how node versions crosswalk with the process.versions.modules
value:
{ '0.6.3': 1,
'0.6.4': 1,
'0.6.5': 1,
'0.6.6': 1,
'0.6.7': 1,
'0.6.8': 1,
'0.6.9': 1,
'0.6.10': 1,
'0.6.11': 1,
'0.6.12': 1,
'0.6.13': 1,
'0.6.14': 1,
'0.6.15': 1,
'0.6.16': 1,
'0.6.17': 1,
'0.6.18': 1,
'0.6.19': 1,
'0.6.20': 1,
'0.6.21': 1,
'0.7.0': 1,
'0.7.1': 1,
'0.7.2': 1,
'0.7.3': 1,
'0.7.4': 1,
'0.7.5': 1,
'0.7.6': 1,
'0.7.7': 1,
'0.7.8': 1,
'0.7.9': 1,
'0.7.10': 1,
'0.7.11': 1,
'0.7.12': 1,
'0.8.0': 1,
'0.8.1': 1,
'0.8.2': 1,
'0.8.3': 1,
'0.8.4': 1,
'0.8.5': 1,
'0.8.6': 1,
'0.8.7': 1,
'0.8.8': 1,
'0.8.9': 1,
'0.8.10': 1,
'0.8.11': 1,
'0.8.12': 1,
'0.8.13': 1,
'0.8.14': 1,
'0.8.15': 1,
'0.8.16': 1,
'0.8.17': 1,
'0.8.18': 1,
'0.8.19': 1,
'0.8.20': 1,
'0.8.21': 1,
'0.8.22': 1,
'0.8.23': 1,
'0.8.24': 1,
'0.8.25': 1,
'0.9.0': 1,
'0.9.1': 10,
'0.9.2': 10,
'0.9.3': 10,
'0.9.4': 10,
'0.9.5': 10,
'0.9.6': 10,
'0.9.7': 10,
'0.9.8': 10,
'0.9.9': 11,
'0.9.10': 11,
'0.9.11': 11,
'0.9.12': 11,
'0.10.0': 11,
'0.10.1': 11,
'0.10.2': 11,
'0.10.3': 11,
'0.10.4': 11,
'0.10.5': 11,
'0.10.6': 11,
'0.10.7': 11,
'0.10.8': 11,
'0.10.9': 11,
'0.10.10': 11,
'0.10.11': 11,
'0.10.12': 11,
'0.10.13': 11,
'0.10.14': 11,
'0.10.15': 11,
'0.10.16': 11,
'0.10.17': 11,
'0.10.18': 11,
'0.11.0': 12,
'0.11.1': 12,
'0.11.2': 12,
'0.11.3': 12,
'0.11.4': 12,
'0.11.5': 12,
'0.11.6': 12,
'0.11.7': 12 }
-
package.json
has aninstall
target that points at./build.js
-
build.js
can be run directly but will also properly handle arguments passed tonpm install
- A small module called
binary_name.js
is used both inbuild.js
andlib/sqlite3.js
to abstract out the details of what a given binary build is called and where it lives. - Binaries are checked for on s3 and if not found a source compile will be the fallback
- sha1 sums are used to validate binaries before testing
- a naive test of whether the module can be required is done and a source compile is the fallback if this fails
- generally this code is quite rough - I plan (@springmeyer) to clean it up eventually into a stand alone module any developer of a C++ addon could use and contribute to.
- A few scripts live in build-util that I use to rebuild binaries on mac and upload them to s3 - this will be cleaned up and integrated into js eventually