If Verdana.ttf can't be loaded, the font file pointed by the
FALLBACK_FONT_PATH environment variable will be loaded.
This fixes width computation errors that can occur with non-latin
characters when running with Docker.
* Add template for secret.json
- Move to faster and lighter Alpine base image
* Update documentation
* Update documentation
* Fix Github token config for secret.json
* Extend env file for Docker runtime configuration
- Update documentation
- Add gh_token for GH personal access token to secret template
* Change http to https in infoSite
* Update .dockerignore
* Update .gitignore
* Update dockerignore
* Remove ENV directive from Dockerfile
- Environment is needed at runtime, not at buildtime
* Docker: contain secret.json in private/
- Incorporates fix from 7c8b0e3d
* Use localhost in example env
* Use baseUrl in GitHub redirect
* Move GH personal token retrieval up
- To remove duplicate Promise.then()
* Typo in shields.example.env
Reorg of the tests: move them just alongside their code. The principle relates to grouping by coupling, not by function and is established in best-practice documents (e.g. https://github.com/focusaurus/express_code_structure#underlying-principles-and-motivations), despite its break from the tradition of a separate `test/` tree. All of today's tools can handle tests spread through the repository.
There are some good, if subtle consequences of this change:
- Since files are close at hand, friction is reduced at development time, which encourages that new tests are written to cover new behaviors.
- It's easier to find the tests that cover a particular piece of functionality.
- It's easier to see which code has tests and which doesn't.
While working on some tests, I was having a tricky problem in a test suite. Eventually I tracked it down to an interaction between tests. I suspected the test library, but once I tried to make an isolated test case, I realized the test library was working fine. It turns out it was the server’s request cache. The fix is to clear the cache between tests.
Not needed for this PR, though I’m adding it to this branch because it conflicts with this change.
Running the server in process is necessary for the mock to work. This is an approach I’ve taken in the past. I experimented with this setup quite a bit when I was playing around with a test suite, and it seemed to work well enough. Setting `process.argv` is a admitedly a bit gross, though a cleaner approach would require more involved changes to `server.js`.
Given the chunks coming from imagemagick are getting stored memory and
then tucked into a cache, this function could as easily return a buffer
via callback. Streaming is just making it more complex. (And trickier to
test!)
In ef1a5159, the switch to using imagemagick made a faulty use of the library by
listening for an 'end' event that is never raised. As a result, the cache was
never populated.
In d985f81f, a fix that takes care of the fact that the previously mentioned
dead code relies on a non-existent variable caused it to kill the server when a
raster badge is requested twice, as what it stored in the cache was the pipe
transmitting chunks, not the chunks themselves, and the pipe (a Socket object)
cannot be subsequently sent through a pipe. The following error occured instead:
events.js:163
throw er; // Unhandled 'error' event
^
TypeError: Invalid non-string/buffer chunk
at chunkInvalid (_stream_readable.js:395:10)
at readableAddChunk (_stream_readable.js:150:12)
at DataStream.Readable.push (_stream_readable.js:136:10)
at DataStream._read (/home/m/shields/lib/svg-to-img.js:45:21)
at DataStream.Readable.read (_stream_readable.js:350:10)
at resume_ (_stream_readable.js:739:12)
at _combinedTickCallback (internal/process/next_tick.js:80:11)
at process._tickDomainCallback (internal/process/next_tick.js:128:9)
The following warning is emitted by Node.js:
> DeprecationWarning: Using Buffer without `new` will soon stop working. Use `new
> Buffer()`, or preferably `Buffer.from()`, `Buffer.allocUnsafe()` or
> `Buffer.alloc()` instead.
This patch removes this warning.
The old LRU implementation stored a list's indices to reference items in that
list, but deletions from the list made indices point to the wrong slot.
Functionally, it meant that deleted slots were not guaranteed to be the oldest
slot.
Using a linked list fixes that.
Christophe BliardandPaul Melnikow
Thaddee Tyl
Adriaan GroenenboomandDaniel Lo Nigro
anatoly techtonikandPaul Melnikow
Paul MelnikowandThaddée Tyl
Frode KnutsenandPaul Melnikow