That's because insertBy' relies on insertUnique being done atomically, which requires ON CONFLICT support, a feature that came with PostgreSQL 9.5.
5.2 KiB
Vervis is still in early development and the build process gets updates once in a while, but this file tries to keep up and list the latest instructions for running a Vervis instance.
At the time of writing, you can get a running Vervis instance if you follow the steps below.
(1) System libraries
Install dependency library development packages. It's very likely you already have them all installed, and if you're missing some, the build process will inform you. But it's still nice to have a list here. The list below isn't a complete list, it's just libraries that people have found missing while building, and let me know.
- PostgreSQL client library
- ZLib
- libssl
On Debian based distros, installation can be done like this:
$ sudo apt install libpq-dev zlib1g-dev libssl-dev
(2) The Stack build tool
Install stack. To install stack, go to its website and follow the instructions.
(3) Version control systems Darcs and Git
Install Darcs. You can grab it from your distro, e.g.:
$ sudo apt install darcs
If you're going to create a Git repository on Vervis, you'll need Git too, you can install it from a distro package too, e.g.:
$ sudo apt install git
(4) The Vervis source code
Clone the Vervis repo:
$ darcs clone https://dev.angeley.es/s/fr33domlover/r/vervis
$ cd vervis
Clone dependency libraries:
$ ./update-deps.sh
(5) Configuration and database
Generate a new SSH key with a blank password:
$ ssh-keygen -t rsa -f config/ssh-host-key
Install PostgreSQL. You'll need the server and the client library development files. Note that PostgreSQL needs to be at least version 9.5.
$ sudo apt install postgresql libpq-dev
Switch to postgres
system user:
$ sudo su - postgres
Create a PostgreSQL user.
With password:
$ createuser --no-createdb --no-createrole --no-superuser --encrypted --pwprompt vervis
No password (if you run Vervis as a user by the same name as the DB user):
$ createuser --no-createdb --no-createrole --no-superuser vervis
Create a PostgreSQL database:
$ createdb --encoding=UTF8 --owner=vervis vervis
Update the settings to specify correct database connection details and other settings.
$ cp config/settings-default.yaml config/settings.yml
$ vim config/settings.yml
(6) GHC Haskell compiler
stack
can automatically install the correct GHC version for you, in an
isolated location that doesn't conflict with any system packages. Unless you
have a specific reason to get GHC in some other way, getting it through stack
is recommended.
If you'd like to install GHC manually (from a distro package, from a PPA,
etc.), this is the time to do so. And I trust you to arrange things such that
stack
uses your manually download GHC. Otherwise, simply proceed to the next
step.
(7) Build Vervis
Build. This will also automatically install GHC.
$ stack build
(8) Development and deployment
To update your local clone of Vervis, run:
$ darcs pull
$ ./update-deps.sh
$ stack build
For convenience, at least on actual deployments, you may wish to run the Vervis SSH server on port 22, so that people don't have to specify a custom port. For that to work, the user that runs the Vervis server needs to get permission to bind to ports below 1024. There are several ways to do that. One of them is to use file capabilities to give the Vervis executable the permission to bind to such ports (if you prefer not to trust the code, try one of the other methods, such as sudo):
$ sudo setcap CAP_NET_BIND_SERVICE=+ep `stack exec which vervis`
Vervis uses various key files for cryptography and other data generation, and once these key files are created, they shouldn't change. For some of them, it's critical they don't change, because some usage or interpretation of data in the PostgreSQL database depends on them. For this reason, by default, key file loading happens as follows: When Vervis runs for the first time (this is checked by detecting that the database is empty, no tables), it generates and writes key files, and it's an error if any of them already exist. Otherwise, on the next time(s) Vervis runs, it requires all key files to exist, and an error is raised if any are missing.
If you're running Vervis for the first time, i.e. the database is still empty, and for some reason you'd like Vervis to load some existing key files, while generating the rest, run this:
$ touch _keyfile_import_existing
Run.
$ stack exec vervis
When you update Vervis to a newer version, it's possible the software now uses some new key files, and Vervis will raise an error about those key files missing in the filesystem. You can ask Vervis to generate missing files, and load the rest as usual:
$ touch _keyfile_write_missing
$ stack exec vervis
Browse to http://localhost:3000
and have fun!
yesod devel
is another way to run the application, useful for rapid
development, but I haven't been using it and I'm not sure it works, possibly I
broke something along the way. But feel free to try!
I have a little script for deploying Vervis on my server. I just haven't published it yet and haven't explained how it works. If you're interested, ask me about it, and it will motivate me to write about it sooner :)