Firefox - Extension to install the latest version

[coreplayer2]

Firefox v38.0.1 is available

Finally a version where flash v11.x works effortlessly

Use firefox-latest.tcz extension to fetch and create the latest firefox-official.tcz,
simply reboot to install.

Consider deleting the 'firefox-official.tcz.md5.txt' file to prevent Apps > Update from downgrading to an older version when you next run Apps > Update

[Misalf]

Great!
Effortlessly for the update script or for Firefox itself?
If the latter, what effort was needed before?
I use the getFlash11 extension regularly to update Flash and it just-works in my personal custom Firefox extension.

[coreplayer2]

the latter,   it seem there was always some occasional minor glitch which required a restart to make flash work...   I'm pleased to find in v38.0.1 no such glitches


The update extension is working great also


However I've not tested yet the phenomena where Flash fails to load when there is no sound module loaded, this is not a firefox issue though..

[coreplayer2]

Update!!

With the help of many testers, thanks guys for your valuable feedback.   We have been updating this extension in stages.

Stage 1,   (completed)
Updates the architecture detection to automatically create Firefox for
i)   x86 (tc-32bit),
ii)  x64 (tc-32bit user-space apps on x86_64 kernel) and
iii) corepure64 (pure x86_64) configurations.

Stage 2,   (completed)
Updated to provide Firefox with locale support (complete language support as provided by Mozilla).
Detection of local language support is based on the "lang=xxxx" bootcode which modifies the download script resulting in creation of a locale specific, latest Firefox extension

Stage 3,   (in progress)
Support for .mar incremental upgrades for folks who are bandwidth challenged (if I can get this to work as advertised).


Other fixes
1. detection of free-space remaining,   (completed)
Which now reports free space in MB's available (instead of %).  This feature also limits firefox download and extension creation with less than 200MB remaining.  (fixed)

2. timeout menu's,   (in progress)
Will try to replace time-out menus with command line arguments.  the idea behind this is to assist in efficient uninterrupted extension build without missing the menu objective.

3, Name changes,    (in progress)
With the current forum discussion of older unsafe Firefox versions in repo's, perhaps it's time to revisit a name change for the created firefox extension.

a, I'm thinking simply firefox.tcz (replacing firefox-official.tcz). 

b, Perhaps also take the opportunity to submit and maintain the current firefox-ESR to the repo, this being the latest Extended Support Release.  long lasting secure revision for the repo, with less frequent updates is easier to maintain..?

c, Also maybe appropriate to change the name of this extension with an appropriate description.
I'm thinking "firefox-getLatest.tcz"  in keeping with the creation of a latest extension as opposed to firefox-latest.tcz which implies is the latest version..?

4, some minor cleanup..



thoughts..??


I've attached here the current script which incorporates the above completed changes, extension will be updated soon.

[coreplayer2]

Progress Update!!

New firefox-getLatest.tcz  (submitted for all x86 and x86_64 repos) which
creates a new firefox.tcz (extension) with locale support, more details below.

Updates to architecture detection for creation of firefox.tcz for  (completed)
i)   core x86 (tc-32bit),
ii)  core64 x64 (tc-32bit user-space apps on x86_64 kernel) and
iii) corepure64 (pure x86_64) configurations.


Locale support  (completed)
locale support for Firefox; complete language support as provided by Mozilla.     
Detection of local language support is based on the "lang=xxxx" bootcode which modifies the
download script resulting in creation of a locale specific, latest Firefox extension


Support for .mar incremental upgrades for folks who are bandwidth challenged (if I can get this to work as advertised).
 (in progress)


Other fixes 
1. detection of free-space remaining,  (completed)
Which now reports free space in MB's available (instead of %).  This feature also limits Firefox download and extension creation with less than 200MB remaining.


2. timeout menu's, (completed)
Replace time-out menus with command line arguments.  the idea behind this is to assist in uninterrupted extension build process


3, Name changes,   (completed)
With the current forum discussion of older unsafe Firefox versions in repo's, perhaps it's also time to revisit a name change for the created firefox extension.

a, Now simply firefox.tcz (replacing firefox-official.tcz).  (completed)

c, Also changed the name of this extension with an appropriate description.
"firefox-getLatest.tcz"  in keeping with the creation of a latest extension as opposed to firefox-latest.tcz which implies is the latest version  (completed)

d, changed permissions of created extension in keeping with tc guidelines

4, some minor cleanup..


Note: the new firefox-getLatest.tcz  replaces
the depreciated firefox-latest.tcz currently in repo
Additionally the newly created Firefox.tcz extension
replaces the depreciated local firefox-official.tcz extension

[coreplayer2]

I've attached the latest script used in the new firefox-getlatest.tcz (extension) here

[Lee]

I've attached the latest script used in the new firefox-getlatest.tcz (extension) here
* firefox-getLatest.sh (14.3 kB - downloaded 3 times.)

I tried out this script under Core 4.7.7 / x86 and it worked except for one instance where the target of wget didn't explicitly start with either "http://" nor "ftp://".  Once I changed

Code: [Select]

wget -q $latestVER
to

Code: [Select]

wget -q http://$latestVER
it ran and produced the extension as expected.

Very nice!

It is, however, an  annoyance that the script goes so far as to add the new extension to onboot.lst.

(FWIW, its also annoys me that tce-load -wi adds things to onboot.lst.  Making an extension load onboot should be available but should be an "opt-in" choice, not an "opt-out" choice.)

[gerald_clark]

Then don't do tce-load -iw.  Instead do
tce-load -w
tce-load -i

[yiyus]

To be honest, the options are quite confusing.

When you want a new package, you can use -wi if you want to add it to onboot, -wo if you want to add it to ondemand, or -wil if you just want to install it for this session. If you only want to download it, you can use -w. Then, later you can install with -i, but the result you get is not like running -wi, it is like running -wil.

I find this not only counter intuitive, but also inconvenient, because I usually just want to try the package (so I use -wil) and decide later if I want to load it on boot, but I have to edit the onboot file for that, tce-load knows how to do it  but it does not handle this case.

Wouldn't it be more natural to have a flag to download, a flag to install, and a flag to add to onboot or ondemmand and combine these flags? I guess it would be hard to find the right moment to make a change like this, but I hope it eventually gets fixed.

(Sorry for the off-topic. The firefox extension looks good, but it has not hit the repos near to me yet).

[coreplayer2]

I tried out this script under Core 4.7.7 / x86 and it worked except for one instance where the target of wget didn't explicitly start with either "http://" nor "ftp://".  Once I changed

Code: [Select]

wget -q $latestVERto

Code: [Select]

wget -q http://$latestVERit ran and produced the extension as expected.

Very nice!

Lee, I'm a little baffled by this because the address to download the web page with current release information always begins ftp.

The base address defined earlier in the script is always this, which includes ftp.

Code: [Select]

latestVER="ftp.mozilla.org/pub/mozilla.org/firefox/releases/latest/linux-i686/en-US"
this address variable is then modified during the architecture check and then again during a locale check

Code: [Select]

latestVER=${latestVER/linux-i686/linux-x86_64}parameter substitution replaces arch (and locale). therefore (assuming en_US locale, the result can only be either
ftp.mozilla.org/pub/mozilla.org/firefox/releases/latest/linux-i686/en-US , or
ftp.mozilla.org/pub/mozilla.org/firefox/releases/latest/linux-x86_64/en-US



Interestingly this is not related to the Firefox archive download and placing http in front an ftp address should have failed to determine the current release version. 
so I think there is possibility the script failed because of some other reason,  perhaps during one of the intentional connection checks.  Then when you tried again the connection quality had improved.. 

Please try again after removing the http://  prefix, I'm anxious to hear the result

[curaga]

@coreplayer2, older busybox wget always required http:// or ftp:// in urls. I submitted a patch for that some time ago, where it adds http:// if no protocol prefix exists. 4.x must be before that.

[yiyus]

It is, however, an  annoyance that the script goes so far as to add the new extension to onboot.lst.

After a second thought, I think there is an easy fix for this. Just change line 550 to:

Code: [Select]

grep -q firefox-getLatest.tcz ${tce_dir}/$bootlist && grep -q firefox.tcz ${tce_dir}/$bootlist || echo "firefox.tcz" >> ${tce_dir}/$bootlist

(or maybe it would be even better to do a s/firefox-getLatest.tcz/firefoz.tcz/ substitution)

Then, firefox.tcz will be added to the onboot list only if firefox-getLatest was in that list, which I think is more polite than always adding firefox.tcz.

[coreplayer2]

@curaga    ok I hear you now..    We need to test for wget version ?? 



@lee   sorry I misunderstood  (I think)


Will look at a resolution to this asap

[curaga]

No need to test if you always include the protocol.

[coreplayer2]

I tried out this script under Core 4.7.7 / x86 and it worked except for one instance where the target of wget didn't explicitly start with either "http://" nor "ftp://".  Once I changed

Code: [Select]

wget -q $latestVER
to

Code: [Select]

wget -q http://$latestVER
it ran and produced the extension as expected.

Very nice!

OK Lee, I understand the problem now.  I  assume the one instance where it failed was on tc-4.x ? which the script has not been tested on otherwise I would have probably caught the issue.  has been tested on tc-5.x and tc-6.x only, in any case busybox version of wget is forced.   I was wrong with the protocol prefix so have added http:// to latestVER="ftp.mozilla.org/pub/mozilla.org/firefox/releases/latest/linux-i686/en-US" 
I'll have to look for an old tc-4.x to test on.  Thanks for pointing this out.

code now reads

Code: [Select]

latestVER="http://ftp.mozilla.org/pub/mozilla.org/firefox/releases/latest/linux-i686/en-US"
this also gave me another idea  (more later..)

It is, however, an  annoyance that the script goes so far as to add the new extension to onboot.lst.

Isn't this the default action of APPS?   
I'm thinking that anyone wanting to download or update a commonly used application would want the browser available each time they boot.. am I wrong in that assumption..??
If I'm wrong with that assumption, then the philosophy behind Apps is also wrong..?
I don't honestly know where to go with this..  I think the consensus would be to have a commonly used application available on every boot.

It's an easy fix but my fear is it would confuse many folks