Back to home page

DOS ain't dead

Forum index page

Log in | Register

Search:  
Back to the forum
Board view  Mix view

FlWriter version 1.0 released (Announce)

posted by Arjay, 21.03.2012, 00:16
(edited by Arjay on 21.03.2012, 08:04)

> I thought hmmm better download via wget,
 Ok successfully d/l via wget. First thoughts re v1.0. Re size I stand by my comments above.

1)Fonts - you can halve your distribution by cutting down on fonts. I'd stick the main "common" fonts, e.g. ariel, courier, times into either a main.zip or mainfont.zip and then stick less common fonts e.g. dejas, vera, libmo etc in xtrafont.zip

2)font variations - ariela, arielb, arieli etc. Ok variations on the same basic font. Does the main distribution need so many variations? When I browse the font directory (on this nix box) I can see that each font has 4-5 variations where the variations are basically angle of characters and level of bold. As an end user how often would I ever use say arialz.ttf ? (right angled slant with bold) vs say arielb.ttf (bold left angled slant). I think I'd probably use arielz.ttf instead of either. I don't know but the choice seems large (which is good) but certainly I'd expect someone to use say arielz.ttf (as ariel is more popular) instead of say libmobi.ttf which is a less popular font.

There are various reports of most popular fonts - so if I was packaging this up I'd use something like that as a guide of what to include in a main.zip vs what to exclude and put into a xtrafont.zip

> > I used UPX with the executables and then ZIP with the package. Therefore
> > you do not gain much with a different compression program.
3)On one of my projects I once spent ages looking at compressing some data, trying different compressors/techniques before it then dawned on me that instead I could just avoid putting in the data in the first place. So with that in mind and building on my fonts comments, is there any data that is duplicated, e.g. for example DJ's stub.EXE which upx doesn't compress. An installer could for example programmatically copy a header which is stored once onto the top of various files which share that header.
EXE files is a bit of a bad example as harder to do (but not impossible) but hopefully you can see what I mean by that comment.

4)contools - ok so only 370k - a good example of something that could be separate download.

5) Main EXE's * 6 total = 4.6meg - ouch. Quick look at flwriter.exe after decompression. Quick thoughts:
strings - are there any common strings between EXE's that could be moved to external config files which would also aid translators.
bitmaps - uncompressed gfx. again any common gfx between exe's that could be moved externally to a single file.
redundant code present - e.g. I spotted a multiple re-occurances of NOP's 90h,90h,90h repeated patterns of code routines that re-occur multiple times.
[paging through EXE I got bored...]

I'd suggest running a strings util (e.g. strings.exe) over the each exe's pumping the output into a text file per EXE and looking to see what's common or redundant. Are you using all the strip functions of the compiler?

6) printer drivers - not sure what these take up but would suggest if large to include say an epson fx80 type driver, basic HP laser and have the others as optional (if need be) since most printers tend to support fallback to them.


Still really good project, so thank you again. I hope you see this as constructive feedback which is how it is intended. I realise though that it is possible to spend hours just on size alone and that there is a trade off but I do think this is an area that as others have said lets this great project down at the moment. There is just too much data in the distribution.

Post reply
 

Complete thread:

Back to the forum
Board view  Mix view
22778 Postings in 2122 Threads, 402 registered users (0 online)
DOS ain't dead | Admin contact
RSS Feed
powered by my little forum