develooper Front page | perl.perl6.users | Postings from August 2022

Re: BEGIN {} question

Thread Previous | Thread Next
ToddAndMargo via perl6-users
August 29, 2022 17:31
Re: BEGIN {} question
Message ID:
On 8/29/22 08:41, Tom Browder wrote:
> On Mon, Aug 29, 2022 at 10:29 AM ToddAndMargo via perl6-users 
> < <>> wrote:
> ...
>  > Does the compiler make a full pass through
>  > the code before firing off the BEGIN routine
> NO.
> And I think you may be surprised how much speedup you may get by using 
> the precompiled-module "trick" for most of your 11,000-line program.
> -Tom

Hi Tom,

The .precomp workaround was never in question!

But there are tines when it is impractical.

Most of the programs I have written for customers
run in the background (rakuw) or at night when
no one cares how long they take to start.

Other programs are to remove mistakes from barcode
data when routed to a bar code program.  The
customer inputs data into their Point of Sale
software that freaks out the bar code printing
program, so I correct it on the fly.  It
is a lot easier than going through their entire
database and having the remote `'` for feet and
`"` for inches in a CSV file sent to the
barcode program (comma-separated value).  Not
to mention it is impossible to get the users
to stop making those mistakes

I also have written my own Dynamic DNS (Domain
Name Server) work around for folks that I
have installed RDP (remote Desktop Protocol) on
and that are using floating WAN (Wide Area
Network) addresses.  But it is pretty simple
and starts quick enough.

Now for where .precomp is not practical.  It is
those projects were the saying "the software is
never finished" applies.   Where I am loath to
start/recommend projects that the customer
expects a reasonable response time to start.
So far, I have been lucky.

For those programs that do not require a lot of
maintenance, the .precomp work around is reasonable.

Primarily where the response time drives me INSANE
is my software updates program.   This program goes
out and checks for new updates for programs I support.
If a new one exists, it downloads.  I carry these
programs with me to customer sites on a read only
flash drive so as to not transfer viruses between
customers.  So far it is about 70 programs.

Because Web Developers are always working on their
sites, I am constantly having to figure out
what is wrong.  "The software is never finished".
So .precomp does not help.

What I have done though is to configure the program
such that if I put the sub name of the target program
on the run line, my program will only run that section
rather that waiting 5 minutes to go through them all.
And if my program finds something on the run line it,
it also triggers my extensive debugging.

And it does take a lot of iterations to debug my regex's.
And the wait time (17 seconds) drives me NUTS.
Sometimes, the "source" code on Firefox's development
tools dos not show hidden character in their pages and
that creates a nightmare too.  I spurt the web page
to a file and then see what the heck is wrong.
I am getting pretty good at it though.  "Git" and
"Brave" (Browser) are interesting to download from.

So lots and lots of compiling that .precomp does not
help me with.

More information that your wanted.  Sorry.


Thread Previous | Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About