294543 - Reporter Doesn't get the Correct Product Name/Version

Status: RESOLVED FIXED

(Other Applications Graveyard :: Reporter, defect)

Description

[Robert Accettura [:raccettura]]

Thanks to bug 274928, we now need to use
general.useragent.extra.firefox

They pulled the carpet out from under us by nixing
window.navigator.vendor+'/'+window.navigator.vendorSub

Patch coming

Comment 1

[Robert Accettura [:raccettura]]

Attachment: Patch

Patch

I'm not quite ready to drop 1.0 support yet.  Still may come in handy, so I'm
leaving that.

Comment 2

[Asa Dotzler [:asa]]

Comment on attachment 183862 [details] [diff] [review]
Patch

a=asa pending review.

Comment 3

[Mike Connor [:mconnor]]

Comment on attachment 183862 [details] [diff] [review]
Patch

r=me, but use Unknown or something like that is what we want, less jarring in
results if that case ever gets hit.

Comment 4

[Robert Accettura [:raccettura]]

Checking in extensions/reporter/resources/content/reporter/reportWizard.js;
/cvsroot/mozilla/extensions/reporter/resources/content/reporter/reportWizard.js,v
 <--  reportWizard.js
new revision: 1.13; previous revision: 1.12
done

Comment 5

[Robert Accettura [:raccettura]]

We aren't done here, we get "unknown" way to often to be satisfied.  I need a
bonified way to get product/version.

Comment 6

[James Ross]

The first block of code at the URL below gets the info you likely want:
http://lxr.mozilla.org/mozilla/source/extensions/irc/js/lib/message-manager.js#54

Note this only works in FF 1.1 and above, but I take it that's more than fine
for this.

As for Mozilla, not so sure on the "best" way...

Comment 7

[James Ross]

Actually, if you only want the product /name/, the branding code below it will
be fine for suite as well - but it wont give you a version. :(

Comment 8

[Robert Accettura [:raccettura]]

Version is critical for any layout use.  So that we can track regressions
between what effects 1.1 users, and say 1.5 users.

Comment 9

[James Ross]

I guessed it probably was, but this bug is therefore incorrectly summaried. :P

Comment 10

[Boris Zbarsky [:bzbarsky]]

The Gecko version is the "rv:..." part of the UA string, no?  That's the part
you need for layout issues.

Comment 11

[James Ross]

Makes me wonder why nsIXULAppInfo exposes the 2005nnnnnn number, and not the rv:
one.

Comment 12

[Asa Dotzler [:asa]]

darin or bsmedberg, do either of you know how we can reliably and easily get
accurate product/version information from our app?

Comment 13

[Christian :Biesinger (don't email me, ping me on IRC)]

chrome://global/content/build.dtd has the build id, in case you need it (may be
seamonkey only)

Comment 14

[Robert Kaiser]

(In reply to comment #12)
> darin or bsmedberg, do either of you know how we can reliably and easily get
> accurate product/version information from our app?

I think best is to query nsIXULAppInfo for it, and if that fails, try to do some
hackery that gets us the Mozilla suite version info (basically name="Mozilla",
and I hope we're able to get the Gecko version somewhere - I guess the same
place that is used for assembling the UA string).
We currently try to get something working for future suite (SeaMonkey) in bug
294943 and it looks like we can go for a static JS implementation of
nsIXULAppInfo, so above way would work nicely with what is to come from our side
(as we just won't go into the failover then).

Comment 15

[Benjamin Smedberg]

KaiRo's plan sounds good to me.

Comment 16

[Robert Accettura [:raccettura]]

(In reply to comment #14)
> 
> I think best is to query nsIXULAppInfo for it, and if that fails, try to do some
> hackery that gets us the Mozilla suite version info (basically name="Mozilla",
> and I hope we're able to get the Gecko version somewhere - I guess the same
> place that is used for assembling the UA string).
> We currently try to get something working for future suite (SeaMonkey) in bug
> 294943 and it looks like we can go for a static JS implementation of
> nsIXULAppInfo, so above way would work nicely with what is to come from our side
> (as we just won't go into the failover then).

A few issues here:

1.  Detecting SeaMonkey isn't the problem.  AFAIK we have no SeaMonkey isn't
effected at all.  We can do that already using:
  if ('nsIChromeRegistrySea' in Components.interfaces) {}

2.  The #1 issue is getting the Firefox release version # in a way that won't die.

3.  Gecko version number is a separate issue.  When someone has a way of getting
that reliably (frozen method), open a bug against me with info on that, and
we'll add it in.  reporter is designed to handle it when it, we just don't use it.


Myk donated the following via a blog comment, but we still don't get the version
number (apparantly that's the worst issue):
var gAppName;
function getAppName() {
    if (gAppName)
        return gAppName;

    try {
        // Aviary 1.1 apps store application meta-data in the app info component.
        var appInfo =
          Components
            .classes["@mozilla.org/xre/app-info;1"]
              .getService(Components.interfaces.nsIXULAppInfo);
        if (appInfo && appInfo.ID ==
"{3550f703-e582-4d05-9a08-453d09bdfdc6}") {
            gAppName = "Thunderbird";
            debug("determined application to be Thunderbird via nsIXULAppInfo");
            return gAppName;
        }
        debug("could not determine application via nsIXULAppInfo: " + appInfo);
    }
    catch(e) {
        debug(e);
    }
    
    try {
        // Aviary 1.0 apps store application meta-data in a preference.
        var id =
          Components
            .classes["@mozilla.org/preferences-service;1"]
              .getService(Components.interfaces.nsIPrefService)
                .getBranch("app.")
                  .getCharPref("id");
        if (id && id == "{3550f703-e582-4d05-9a08-453d09bdfdc6}") {
            gAppName = "Thunderbird";
            debug("determined application to be Thunderbird via app.id pref");
            return gAppName;
        }
        debug("could not determine application via app.id pref: " + id);
    }
    catch(e) {
        debug(e);
    }

    // Non-aviary apps (f.e. Seamonkey) don't store application meta-data.
    debug("could not determine application; assuming Seamonkey");
    gAppName = "Seamonkey";
    return gAppName;

Comment 17

[James Ross]

(In reply to comment #16)
Did you miss the code I linked to in comment 6? nsIXULAppInfo returns not only
the application name (you can use the ID if you like too), but the version,
*and* the gecko ID! (app.geckoBuildID)

That still leaves suite, but work is afoot in bug 294943 to add nsIXULAppInfo to it.

Comment 18

[Robert Accettura [:raccettura]]

(In reply to comment #17)
> (In reply to comment #16)
> Did you miss the code I linked to in comment 6? nsIXULAppInfo returns not only
> the application name (you can use the ID if you like too), but the version,
> *and* the gecko ID! (app.geckoBuildID)

Was away for a few hours, and didn't reread the bug from the very beginning,
just the new stuff.  Yea, I see that...

> 
> That still leaves suite, but work is afoot in bug 294943 to add nsIXULAppInfo
to it.

I'm still not really crazy about using something unfrozen.  I'm leaning towards
avoiding it until it freezes.  

Right now I just want to get the app version # in Firefox 1.0 and later, and
SeaMonkey 1.8 and later.

I care about Gecko... just not at the moment.

Comment 19

[Robert Accettura [:raccettura]]

Attachment: Patch v2

Comment 20

[Robert Accettura [:raccettura]]

From what I can tell, the new patch should correctly find:

Firefox 1.0 and later
SeaMonkey 1.8 and later

Comment 21

[Christopher Aillon (sabbatical, not receiving bugmail)]

Caveat: vendors are supposed to use "vendor" stuff.  So, you will get things
like "Fedora/1.0.4-1.3.1" etc for Fedora builds.  That was the whole point of
the patch to bug 274028.  So you may wish to query the extra.firefox stuff
first, then if that doesn't exist checking the vendor stuff, though that still
may not be as reliable as you like for seamonkey.

Comment 22

[Robert Accettura [:raccettura]]

Attachment: Patch v3

Comment 23

[Benjamin Smedberg]

Comment on attachment 184207 [details] [diff] [review]
Patch v3

Same problem as with other patch: no need for "if (appInfo)" after getService.

Comment 24

[Robert Accettura [:raccettura]]

Attachment: Patch v3.1

Address review comment

Comment 25

[Asa Dotzler [:asa]]

Comment on attachment 184214 [details] [diff] [review]
Patch v3.1

a=asa

Comment 26

[Robert Accettura [:raccettura]]

Checking in extensions/reporter/resources/content/reporter/reportWizard.js;
/cvsroot/mozilla/extensions/reporter/resources/content/reporter/reportWizard.js,v
 <--  reportWizard.js
new revision: 1.16; previous revision: 1.15
done