NYCPHP Meetup

[David Krings]

Hi!

First of all, thanks John for your reply. Yes, I do use the WAMPP stack, which 
in several tests happened to come up on top in regards to performance compared 
to LAMPP and IIS. I am not surprised that it performs better than IIS, but was 
suprised about the LAMPP. Although, performance wasn't the reason I chose it. 
I have a W2k Pro box with server SCSI drives that is rock solid and runs for 
months without problems. My record is 9 months without reboot, unheard of in 
the Windows world. And if there is a reboot, it is because Patch Tuesday.

I did dig a bit deeper into gettext and the tools that are available on *nix 
are mostly ported to doze as well. I got tired of the resulting .po files to 
be littered with all kinds of string garbage or less than half the strings no 
matter what I did.
Plan B was writing two simple functions, one that reads in a specified ini 
file into an array, and when a section was specified, returns only that 
section while ditching the remains. The second function accepts a key and a 
default string (the current english string) and has to get passed on the array 
with the strings. I crafted an ini file with 300 entries and strings that each 
are about 200 chars. In my first test I had PHP barf up the entire array, then 
in a second test pick out random strings. I braced for the worst, but turns 
out that this is really fast.
My plan is to create sections for each script that generates output, read in 
the file, then based on the name of the script keep the required section and 
throw the rest away (hoping that PHP releases the memory used). Overall, the 
two functions are less than a dozen lines of code.
That approach has two disadvantages. The code once instrumented gets more 
difficult to read, but using some one or two letter name for my function and a 
short array name will condense the calls. It is not as  short as _(), well, I 
could name my function _, but I don't think that is a good idea. Maybe ~() or 
€() isn't taken yet. The other disadvantage is the manual string extraction, 
but I need to rework all echos anyway and given the state of the .po files 
those would need some massive manual cleanup work as well.

My scripts aren't as logic / display separated as they could be, but I do make 
a point in preparing everything first and then focus on output as much as 
possible, but still, it is sometimes within the same script. I go with the ini 
file approach and see how that works out. The tests worked much better than I 
expected. I then can also expand the functions to accept additional parameters 
if needed.


David