Paul M Jones wrote a blog post around PSR-7 implementations and even though his response was zero I was intrigued by packagist results. Since I am maintaining an implementation of PSR-7 UriInterface
, I was curious to see despite his opinion how many different implementations really exists and their quality.
So looking a the numbers of implementations, I thought it would take me hours to go through the list but I was wrong. The packages found can be quickly divided in 4 categories:
- The first category represents cheating developers. The package maintainer just copy/paste a famous implementations (usually Diactoros) and just changed the namespace or aliased it and claimed that it is a complete implementation. Without going into copyright or infringement I’m baffled at how someone could do that, publish it and claim it as his/her work or think that he/she could benefit from it.
- The second category represents cheating packages that advertise themselves as implementing PSR-7 while in reality they just don’t. When you look at the code source you will found no traces whatsoever of PSR-7 except that they claim to provide such feature. Again another strange technique to promote, someone package.
- In the third category you’ll find packages which are really badly written. They do require the PHP-FIG interfaces and implement them. But the code quality is so bad anyone downloading or using them are really doing themselves a bad service.
- Last but not least their are the good, quality tested package like Slim, Diactoros or Guzzle and packages that just require them to work. This does not mean that those packages do not contain bugs or shortcoming but by reading the code and looking at the test suite you directly spot the code quality.
TL;DR
The sheer number of PSR-7 implementations is a false promise. Don’t believe a book by its cover. Whatever a library or a package advertises, first look at its code and try to measure the package quality before using it. Using Composer and being on Packagist is so common nowadays that it should never be taken as a quality measure.
Hi,
Have you gone through the tests of the repo?
Basically the post is about mutability of a stored object via withAttribute .
Not sure if you have noticed Matthew Weier O’Phinney ‘s comment regarding the same : http://paul-m-jones.com/archives/6510#comment-419448
Hi,
My post is not about the mutability aspect of PSR7 but about the quality of packages claiming that they do provide a complete PSR-7 implementations on packagist. Paul question has been answered By Sarah McG and Matthew Weier O’Phinney. And both answers are IMHO valid.