delorie.com/archives/browse.cgi   search  
Mail Archives: djgpp/1998/06/22/02:29:20

Date: Mon, 22 Jun 1998 09:29:11 +0300 (IDT)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: ^Magico^ <bfe2333 AT mail DOT telepac DOT pt>
cc: djgpp AT delorie DOT com
Subject: Re: Int2TeX v1.00 was released
In-Reply-To: <358e8365.2105653@news.telepac.pt>
Message-ID: <Pine.SUN.3.91.980622091241.9130C-100000@is>
MIME-Version: 1.0

On Sun, 21 Jun 1998, ^Magico^ wrote:

> I'm also doing some other thing's that will do a final info file,
> small in space, and quickly to access, and also less memory will be
> used loading the file.

You might consider NOT using --no-split option to makeinfo.  This way, 
it  splits the Info file in several (more than 120, in this case, 
actually ;-) small files, so that both the load time and the memory 
footprint of info.exe are much smaller.  (The full Info file is above 
6MB, and info.exe, in the best traditions of GNU software, reads it all 
into memory).

> By the way, the numbers are every INT being processed (seeing that you
> can see that the processing needs to be faster). When dashes are being
> printed that means that all topics are being processed. The points
> means that data is being transfered!

Btw, the way you print the dashes and the dots is not very nice. The
problem is that you use cprintf to scroll the display [cprintf("\r\n")]. 
This is a bad idea, since what cprintf does is not very efficient.  For 
example, on Windows, when the display is scrolled, my mouse cursor stops 
moving!!  (I really don't understand why you didn't use fprintf to stderr 
instead of conio functions.)

> Resuming, 80% of time it's being losted with topics check and menu
> items consctruction, and other 20% are being used for transfering
> data!

I suggest profiling the program to find hot spots.  Personally, 15 
minutes on a P166 seems like too long, even for 6MB.  For comparison, 
makeinfo needs only 10 minutes to process the same file, and it does a 
lot more (I think).

I also noticed something that might be a bug: not all the functions of 
the interrupts wind up in the converted file.  For example, INT 21h only 
has functions 01 to 24h, both in the menu and in the body.  Seems like 
int2tex stops recording functions of a particular interrupt after it 
finishes the first INTERRUP.x file that includes that interrupt.  I see 
similar problem with INT 2Fh.  Do you see this also, or is it something 
specific to my system?

Oh, and I have two comments about names.  Texinfo is written like shown: 
with a single capital T (the i is in lower-case); and the output file 
should have either .txi or .texi extension, since .tex is reserved for 
TeX files (TeX is an entirely different language).  This might seem like 
nitpicking, but for example some editors would automatically enter TeX 
mode when you load a .tex file.

- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019