PHP Serbia Conference 2019
24 - 26 May 2019
at Belgrade
Talk comments
Excellent presentations! I knew about a few of the APIs you mentioned but learned a few new ones, presented in an engaging way.
One suggestion: bump up the font size on your slides/examples. Seemed like you had enough room to do so most/all of the time, and that would help when presenting on smaller screens.
“Time is an illusion” was great! I remember that as it was yesterday.
I thought that today it will be the same great level.
Clear points, great talk! Thank you!
Nice speech, mostly things we heard before
Great speech, I liked the info I collected
I have no words to describe the awesomeness that just happened!
Very important talk. Makes the younger developers aware of the specificity of PHP.
Thanks, I hate it.
In all seriousness, the material was well-delivered, and serves as a reminder of a few pieces of the language that are still rather counterintuitive. Fortunately some pieces will be cleared up in the next few PHP versions, but this talk underscores "use braces and parentheses for absolutely everything" :)
Great presentation, good content, nice slides.
However, I missed the real advantage of performing Smoke Tests, at least the way it was recommended in the talk.
Let's say it's a (large) eCommerce site with 100,000 products - why would you request all those 100,000 URLs instead of for example just the important site features like checkout, homepage, main categories and a few products (e.g. 5)? If the application is properly tested (unit, integration and acceptance) it is almost impossible that something goes wrong.
Yes, of course, the data is different on production compared to CI where the tests run, but for those few very rare cases there should be proper server monitoring, logging and exception handling in place. If the production data is the problem, is it really the task of CI / CD to run smoke tests against all 100,000 product URLs? Maybe, maybe not. I would have loved to hear more information about that, where the real gain of the Smoke Tests are. It's always a trade-off (among costs and the outcome) and I'm really not sure how much the value here is (for a lot of dev work, maintaining the URL lists, CI server time, waiting time before it gets deployed, etc.).
Otherwise a great talk and thank you for presenting it. It's always great to see how other companies / teams do things.
Well presented.
Was a bit distracting that the presentation wasn't full screen, but that's the biggest nitpick that I have.
BTW, NIST is a US thing, so I imagine that they're advising the NSA rather than GCHQ. Also, apparently the reason for the 16 character password limits seen in a few places is NTLM hashes top out at 16 characters, after which stuff truncates. Bcrypt's limit is 72, at which point it really doesn't matter :)