Why porting Mozilla to Amiga??

[Hammer]

KennyR wrote:

On my 850MHz Duron notebook, it takes just 10 seconds to start up Phoenix (now Firebird) and display my homepage (fetched from remote server via DSL). That's not too shabby, I'd say. Even with IE's being partly loaded already, it takes 9 seconds just to open and display an empty window on the same machine.

Ten seconds is an age, especially when you're just clicking on something on IRC for general interest. My system boots faster than that. IMO a browser taken 10 seconds to load on a UDMA drive with a 850MHz CPU powering it is unforgivable for "just" a browser.

I guess my definition of slow is just different from all the Linux and Windows users in here. How you have the patience for it all I can't guess. Even my 040/25 felt a sharper machine to use.

In my test X86 PC**, Mozila V1.3 takes about 1.5 seconds to load.

Such as system is powered by a Seagate 80Gb 7200 RPM (yields about ~55Mb/s from aHead’s NeoBurn5’s hard drive test) and Microsoft’s UDMA IDE drivers. The motherboard is ASUS nForce II 400 Ultra based (it’s faster on nVidia specific UDMA IDE drivers, but that’s another issue).

I.E 6.0.2600 loads about similar time as with Mozila V1.3. This is on Windows XP Pro-SP1 with all display frills turned on. Hard drive’s throughput speed, IDE drivers and available physical RAM does play critical role.

Mozilla 1.3 is only ~25Mb. Ideally, with a 55Mb/s hard drive the system should be able to load Mozilla about ~0.5 of a second. My test machine load it at ~1.5 seconds due to overheads (e.g. Windows, seek times, network connection checking and 'etc'). A 10-seconds load time is ‘slow’ for an UDMA Hard disk.

At the moment, I don’t have access to KT133A/VIA 686B (MSI built) equipped PCs.  Maybe later.... It can reach ~40Mb/s in a similar NeoBurn test hard disk speed conditions (using an older Seagate 7200 RPM 40Gb drive)).

I’ll probably test Mozilla’s load speed on a Pentium II 400Mhz with 3.2Gb hard disk and 192Mb RAM(loaded with WinXP) (Later, IF I have the time).

[bbrv]

Hey Bill, where have you been?

Read around Hero!

Remember your email to us?

Just Bill

[bhoggett]

@bbrv

Remember your email to us?

Yes, what about it? It's content has nothing to do with this conversation, or indeed anything else. It was also written in reply to your sending me unsolicited e-mail, something I recall you doing on a number of occasions?

I wonder why you have this habit of bringing up what is supposed to be private e-mail in public? (not that there's anything incriminating or even inconsistent in the mail in question. Or even private at the end of the day.)

[mikeymike]

@ Hammer

Mozilla 1.3 is only ~25Mb. Ideally, with a 55Mb/s hard drive the system should be able to load Mozilla about ~0.5 of a second.

Very flawed assumption.  IDE disks may be able to get a maximum throughput of say 55MB/sec, but for loading lots of little files you'll find the throughput drops to maybe a meg a second.  The entirety of Mozilla's app install doesn't need to be loaded into memory however.

People complain about large libaries or executables, but it is far worse to have lots of tiny libraries that all have to be loaded in rather than one big one.

Apps like Firebird (formerly 'Phoenix) and Thunderbird are currently getting their libraries compacted into one for this reason primarily.

The only time you'll ever see your hard disk perform at anywhere near its maximum throughput is when copying internally between platters, and a disk benchmarking at 55MB/sec will probably manage a maximum of 40MB/sec when everything is taken into consideration, and that's when butterflies aren't flapping their wings around Mount Fuji, and when the very large files (over 200MB say) aren't at all fragmented, and neither is space allocated as their destination.

The most important statistic for a hard disk to have quoted would be latency on requests, which doesn't get officially stated, only in independent benchmarking.  That's the statistic you want to pay attention to, if you want app and data files to be loaded in quickly.

[TurboLaban]

trgse wrote:

Qt is NOT FREE!!!!!!!

Apple went to all the trouble of removing Qt from KTHML, why?

Well, Qt for Linux is available in a GPL version.
So, it is possible to port this to any other platform.

There is already a Qt port available for Mac OS X, but Apple would have to pay a lisence fee to Trolltech to use this in their programs.
Instead of doing this, they have created their own library implementing only the Qt classes used by KHTML. They have not really removed the Qt dependencies from khtml.

--
Glenn Hisdal

[Hammer]

Very flawed assumption.

I have already out-lined some reasons why the load time was not .5 seconds. Refer to

"Mozilla 1.3 is only ~25Mb. Ideally, with a 55Mb/s hard drive the system should be able to load Mozilla about ~0.5 of a second. My test machine load it at ~1.5 seconds due to overheads (e.g. Windows, seek times, network connection checking and 'etc').

IDE disks may be able to get a maximum throughput of say 55MB/sec, but for loading lots of little files you'll find the throughput drops to maybe a meg a second.

That's too low for PAL TV resolution A/V work. All I can say it’s capable of doing full TV resolution (PAL) capturing without dropping a frame (e.g. AVI/DiVX/DVD/Mpeg2 via software encoding).

Have you installed a real time hard disk monitor? There are at least 41 objects within the Mozilla1.3 directory.

The entirety of Mozilla's app install doesn't need to be loaded into memory however.

What proportion in regards to "The entirety of Mozilla's app install doesn't need to be loaded into memory"?  

From a free memory of 261Mb down to 244Mb (after loading Mozilla 1.3). I’ll say it’s close to the 25Mb file load. Defragging them into local placement will help with the load times (I haven't applied this yet).
 

The only time you'll ever see your hard disk perform at anywhere near its maximum throughput is when copying internally between platters, and a disk benchmarking at 55MB/sec will probably manage a maximum of 40MB/sec when everything is taken into consideration,

Who said that was HD's maximum throughput? Note that I haven't used the specific nForce IDE drivers yet.  
 

The most important statistic for a hard disk to have quoted would be latency on requests, which doesn't get officially stated, only in independent benchmarking. That's the statistic you want to pay attention to, if you want app and data files to be loaded in quickly.

A practical 1MB/s second throughput would be a crawl e.g. 512MB 'hiberfil.sys' will take awhile to load (i.e. 512 seconds). My old system loads 512MB 'hiberfil.sys' around ~20~25 seconds .

[gary_c]

KennyR wrote:

Ten seconds is an age, especially when you're just clicking on something on IRC for general interest. My system boots faster than that. IMO a browser taken 10 seconds to load on a UDMA drive with a 850MHz CPU powering it is unforgivable for "just" a browser.

Well, I guess maybe my perception has become skewed. I've been using mainly Windows for a couple of years now and in that context the load time seems pretty good. Other apps, like graphic editors, Dreamweaver, OpenOffice.org and even Eudora take quite a bit longer.But I generally just open them once and they're up all day, so load time isn't really that big an issue for me.

I guess my definition of slow is just different from all the Linux and Windows users in here. How you have the patience for it all I can't guess. Even my 040/25 felt a sharper machine to use.

When I was using BeOS almost all the time I would've had a different reaction, too. (BeOS doesn't even have a busy pointer.) Now, as soon as my ram arrives I'll get my Pegasos up and readjust my load time concepts. :-)

-- gary_c

[iamaboringperson]

i would like to see mozilla for the features, and support for different web sites

if you dont like the speed/size use another browser for everything else