/Ken/ said:
> Let me set the stage: I am using SeaMonkey v1.0.1 and have used prior
> versions of it and Mozilla for some time. Recently I downloaded a large
> file (about 12 MBs) on my computer (Win 98 & still using dial up) and
> for some reason the complete 12 Meg file was not saved. When I looked
> at the file it was about 10.5 MB, and of course delivered a message that
> it was corrupt or incomplete when I executed it.
> Since I wanted the file, I went back and started to download it again,
> pointing it to the same directory where the corrupted file resided.
> Although it started to save the file at about the same rate as the prior
> download, eventually it appeared as if it had recognized that some parts
> of the corrupted download were valid and the progress accelerated at a
> faster rate. Now I am not sure just why it was much faster, but it was
> once the second attempt reached a certain point in the download.
> Now my question: Since each new version of Mozilla or SeaMonkey
> requires a complete download rather than just the parts that changed
> from the previous version, would the latest version use any of the prior
> version if I changed the name of the prior version to the current file
> name and downloaded it over it as described above? Now I know some will
> think I was imagining the progress of the second download described
> above, but I assure you I was not. Something somewhere was being used
> from the prior download for the second attempt. It might have been in
> cache or part of the corrupted file.
> What do you think???

In my experience, if the incomplete file remains in cache, retrieving the
same file will indeed take advantage of it, making the d/l appear to be much
faster at the outset.

On Sun, 14 May 2006 08:55:37 -0500, Ken wrote:

> Now my question: Since each new version of Mozilla or SeaMonkey
> requires a complete download rather than just the parts that changed
> from the previous version, would the latest version use any of the prior
> version if I changed the name of the prior version to the current file
> name and downloaded it over it as described above?

Absolutely not. Mozilla, or any download manager + FTP server,
has no way to do binary-level diffs and give you just the changed
sections.

What you're suggesting would only work if new version of Mozilla
were packaged in a way where all the changed information from the
last version were at the end of the distribution file. But
that's not going to be the case.
I can pretty much guarantee you that the first 1000 bytes of the
SM1.0 distribution are not the same as the first 1000 bytes of
the SM1.1 distribution. Which should suggest to you that you
need to download the whole file.

Get yourself a download manager (I use "Internet Download
Manager" v4.something), and set all my downloads to happen in the
middle of the night, when I'm otherwise not online.

A 12mb file would only take ~35 minutes at 50kbps. You could eat
dinner in that time.

> Now I know some will think I was imagining the progress of the second download
> described above, but I assure you I was not. Something somewhere was being used
> from the prior download for the second attempt.

Not at all. It picked up where it left off - same as if you'd
have pressed the "Pause" link in the download manager and
restarted again later.

-sw

Steve Wertz wrote:
> On Sun, 14 May 2006 08:55:37 -0500, Ken wrote:
>
>> Now my question: Since each new version of Mozilla or SeaMonkey
>> requires a complete download rather than just the parts that changed
>> from the previous version, would the latest version use any of the prior
>> version if I changed the name of the prior version to the current file
>> name and downloaded it over it as described above?
>
> Absolutely not. Mozilla, or any download manager + FTP server,
> has no way to do binary-level diffs and give you just the changed
> sections.
>
> What you're suggesting would only work if new version of Mozilla
> were packaged in a way where all the changed information from the
> last version were at the end of the distribution file. But
> that's not going to be the case.
> I can pretty much guarantee you that the first 1000 bytes of the
> SM1.0 distribution are not the same as the first 1000 bytes of
> the SM1.1 distribution. Which should suggest to you that you
> need to download the whole file.
>
> Get yourself a download manager (I use "Internet Download
> Manager" v4.something), and set all my downloads to happen in the
> middle of the night, when I'm otherwise not online.
>
> A 12mb file would only take ~35 minutes at 50kbps. You could eat
> dinner in that time.
>
>> Now I know some will think I was imagining the progress of the second download
>> described above, but I assure you I was not. Something somewhere was being used
>> from the prior download for the second attempt.
>
> Not at all. It picked up where it left off - same as if you'd
> have pressed the "Pause" link in the download manager and
> restarted again later.
>
> -sw

Thanks for ALL replies. Well it was a good idea if it had worked. You
can understand how the thought crossed my mind, since on the second
download attempt the file it downloaded much faster.

Just for curiosities sake, HB suggested he too had experienced the
faster download when part of a previous attempt was still in cache. Is
part of the file being in cache the critical factor, or would it still
help if the corrupted part were in the same destination directory??

And if it residing in cache is the critical part, would turning off the
computer and attempting a download later fail to see what was in cache
from the previous download??