Today I spent 1.5 - 2 hours trying to debug an issue I was having with PHPMailer -- or was it my foreach loop, or perhaps a misspelled variable, or maybe that method name was supposed to start with a capital letter...?
Just kidding. The issue is fixed at this point, and I did at least learn something new today.
But I'm sure that cycle of questions is familiar to you. Questions like those seem to always clot together in the mind when something that is broken obviously shouldn't be -- or should it be? At that moment, who knows?
Today's problem -- I thought -- was caused by switching from PHP 5.3.x to 5.6.x for a specific codebase (see edit below).
What ended up being the culprit was the lack of the keyword global at the top of a function that was using global variables. These variables (two of them, it turned out) were not being set within the function.
Lessons learned today
Don't automatically trust your own code. Even if it worked semi-recently, "semi-recently" isn't "now".
Look into language upgrades more often. Rare issues that occur between language versions do still happen, even if they are rare. Since not all projects use the exact same sets of features, looking for possible breaking changes between versions is definitely important.
Lingering confusion is probably worth looking into. After typing all of this, and knowing that the variables suddenly weren't accessible anymore - and looking into it for a little longer -- I'm still not 100% sure that the language upgrade is what caused this issue; it seems that the
globalkeyword was available back in PHP 4 (maybe someone with more experience knows).
Back to the question:
What have you been stuck on for some time that turned out to be unbelievably simply to fix?
P.S. I realize an upgrade from 5.3.x to 5.6.x is fairly "behind the times", but it happens to be where we are on a VPS we use. I do see that changing in the foreseeable future though.
Also, I apologize if my writing is all over the place. While I do look forward to being able to write more, this was only meant to be a quick little post anyway (and it may have even grown beyond that 😅).
Thanks for stopping by to read!
Edit: regarding #3 above, I'm definitely more confused now... A local dev who's used PHP since 5.2 told me without the global keyword, it will always error out.
Top comments (10)
A procedure I had running was going backwards, and I simply had two swap two parameters around and the whole the worked as intended.
Backwards? Wow, that's definitely a new one on me.
It was a stupid mistake by myself. Instead of it working out like 4 - 3 = 1, it was doing 3 - 4 = -1.
But it took me roughly 3 months to figure out it was a problem, i.e. when it went into production aha. My testing was showing correct values, but that's because I was faking the math correctly.
It was a learning experience :)
That reminds me of that post about things we always google. One of mine always seems to be the param order for the callback passed to
Array.prototype.map().Param order can definitely get you. But I'm now glad to have hints in both Atom and VS Code about what the next param is expected to be -- at least if the function is defined in the code, or is a native member of the language.
I don’t have a nice pneumonic to help but I realized one day that
map,forEach, and a few others all have the same callback Parma order ofvalue,index,arrayI kept flipping
indexandvalue(and forgetting aboutarray, honestly). But that order does make a lot more sense since theindexandarrayaren't needed for everything.Thanks for mentioning that by the way. I guess they would've done that for consistency and ease of use.
We had sidekiq jobs setup to send out emails. We made a few changes to the template one day and then it started.
Some outgoing email had the new email template and some were still using the old template. This baffled everyone on the team, QAs were frustrated too since they could not reproduce it consistently in the QA env. This was supposed to be rolled out the next day, hence a sense of urgency too.
BUG
We had recently setup a new sidekiq jobs server and did not decommission the old one (we planned to use it for another env). The new code was deployed to the new servers. Whenever the old server would pick the job it would send out an email with the old template causing the confusion.
I was writing a unit test for
POSTrequests on an api endpoint and was wondering why no data was being returned. I tried logging the request object and the data expected was being sent back; it was a wild situation that I almost threw the towel in. Apparently, the endpoint was responding with a204status code that literally meantNO CONTENT. That's a funny TIL moment for me.Ok, I ended up deleting a sentence because I couldn't get
strikethroughto work. Turns out I was using the wrong character. 😅Oh yeah, those are my favorites (wrong letter case).