841711 - httpd.js extension in DEBUG=1 profiles is broken in current nightly

Status: RESOLVED FIXED

(Firefox OS Graveyard :: Gaia, defect)

Description

[Julien Wajsberg [:julienw]]

Since about this week, we can't run our Gaia unit tests in current nightly anymore.

STR:
* rm -rf profile
* DEBUG=1 make profile
* <nightly path>/firefox --no-remote -profile ./profile
* open the Error Console

In Error Console I get the following messages :

Warning: WARN addons.xpi: Exception running bootstrap method startup on [e-mail]: [Exception... "Component returned failure code: 0x80520012 (NS_ERROR_FILE_NOT_FOUND) [nsIXPCComponents_Utils.import]"  nsresult: "0x80520012 (NS_ERROR_FILE_NOT_FOUND)"  location: "JS frame :: resource://gre/modules/XPIProvider.jsm -> file:///home/julien/travail/git/gaia/profile/extensions/httpd/bootstrap.js :: startup :: line 13"  data: no]
Source File: resource://gre/modules/XPIProvider.jsm -> file:///home/julien/travail/git/gaia/profile/extensions/httpd/bootstrap.js
Line: 13

Error: Error: No data in the file
Source File: resource://gre/modules/XPIProvider.jsm -> file:///usr/share/mozilla/extensions/%7Bec8030f7-c20a-464f-9b0e-13a3a9e97384%7D/%7Bd10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d%7D/bootstrap.js -> file:///usr/share/mozilla/extensions/%7Bec8030f7-c20a-464f-9b0e-13a3a9e97384%7D/%7Bd10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d%7D/lib/filterStorage.js
Line: 375

Error: Error: No data in the file
Source File: resource://gre/modules/XPIProvider.jsm -> file:///usr/share/mozilla/extensions/%7Bec8030f7-c20a-464f-9b0e-13a3a9e97384%7D/%7Bd10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d%7D/bootstrap.js -> file:///usr/share/mozilla/extensions/%7Bec8030f7-c20a-464f-9b0e-13a3a9e97384%7D/%7Bd10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d%7D/lib/filterStorage.js
Line: 375

Error: Error: No data in the file
Source File: resource://gre/modules/XPIProvider.jsm -> file:///usr/share/mozilla/extensions/%7Bec8030f7-c20a-464f-9b0e-13a3a9e97384%7D/%7Bd10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d%7D/bootstrap.js -> file:///usr/share/mozilla/extensions/%7Bec8030f7-c20a-464f-9b0e-13a3a9e97384%7D/%7Bd10d0bf8-f5b5-c8b4-a8b2-2b9879e08c5d%7D/lib/filterStorage.js
Line: 375

Warning: function withDir does not always return a value
Source File: resource://gre/modules/osfile/osfile_async_worker.js
Line: 175
Source Code:
       return f.call(file);

Warning: reference to undefined property options.bytes
Source File: resource://gre/modules/osfile/osfile_shared_front.jsm
Line: 85

Error: this.docShell is null
Source File: chrome://global/content/bindings/browser.xml
Line: 323

Not sure everything is related to this problem though.

Comment 1

[[:fabrice] Fabrice Desré]

Attachment: patch

sigh...

Comment 2

[Julien Wajsberg [:julienw]]

Comment on attachment 715196 [details] [diff] [review]
patch

This makes the unit tests work in current nightly but I get strange problems in Aurora: 

* the test output is sometimes disturbed (missing new lines for example
* the test-agent crashes at the end

I didn't try on b2g-desktop but I believe there will be the same problem and we need to be able to run the tests on this.

I wonder if this is possible to keep both imports ?

Comment 3

[Julien Wajsberg [:julienw]]

Attachment: quick&dirty patch

Maybe this is possible to do better than this try/catch but this works for me in both a current Nightly and Aurora

Comment 4

[[:fabrice] Fabrice Desré]

Comment on attachment 715433 [details] [diff] [review]
quick&dirty patch

Review of attachment 715433 [details] [diff] [review]:
-----------------------------------------------------------------

::: tools/extensions/httpd/bootstrap.js
@@ +18,1 @@
>  

I would rather do:

try {
  Cu.import('resource:///modules/devtools/dbg-client.jsm');
} catch(e) {
  Cu.import('resource://gre/modules/devtools/dbg-client.jsm');
}

Comment 5

[Julien Wajsberg [:julienw]]

Attachment: patch v3

Load dbg-client.jsm from the correct URI in new Firefox
---
 tools/extensions/httpd/bootstrap.js |    6 ++++++
 1 file changed, 6 insertions(+)

Comment 6

[[:fabrice] Fabrice Desré]

Comment on attachment 715991 [details] [diff] [review]
patch v3

Review of attachment 715991 [details] [diff] [review]:
-----------------------------------------------------------------

Since we are changing stuff in httpd.js, can you bump the maxVersion in install.rdf to 22 instead of 21? thanks!

::: tools/extensions/httpd/bootstrap.js
@@ +10,5 @@
>    const Cu = Components.utils;
>  
>    Cu.import('resource://gre/modules/Services.jsm');
> +  try {
> +    // FF < 21

Nit: It's actually Gecko, not FF

Comment 7

[Julien Wajsberg [:julienw]]

Attachment: patch v4

* Load dbg-client.jsm from the correct URI in new Firefox
* bumped up the version
---
 tools/extensions/httpd/bootstrap.js |    6 ++++++
 tools/extensions/httpd/install.rdf  |    6 +++---
 2 files changed, 9 insertions(+), 3 deletions(-)

(note: I have also automatic space changes in my tree, I removed them from the patch. Do you think it is ok to commit them or should I remove them before pushing ?)

Comment 8

[[:fabrice] Fabrice Desré]

Comment on attachment 716435 [details] [diff] [review]
patch v4

Review of attachment 716435 [details] [diff] [review]:
-----------------------------------------------------------------

r=me with nit addressed. About the whitespace changes, I they are on fully blank lines I don't care. For others, I kind of dislike that this changes the blame.

::: tools/extensions/httpd/install.rdf
@@ +5,5 @@
>  
>    <Description about="urn:mozilla:install-manifest">
>      <em:id>[e-mail]</em:id>
>      <em:name>httpd.js</em:name>
> +    <em:version>0.2</em:version>

Revert that change, since there's no real feature change in the add-on.

Comment 9

[Julien Wajsberg [:julienw]]

Attachment: patch v5

carrying r=fabrice

asking for approval as I'd like to have this in v1-train and it can't be safer.

master: https://github.com/mozilla-b2g/gaia/commit/58b660c1212461f5d310a9800e16c1bcf097031b

Comment 10

[Lukas Blakk [:lsblakk] use ?needinfo]

Comment on attachment 716715 [details] [diff] [review]
patch v5

approving low risk uplift to v1-train.

Comment 11

[Julien Wajsberg [:julienw]]

I think this broke it for running tests in Aurora and earlier Firefoxes (whereas my patch v4 worked correctly because I specifically tested it). Filed Bug 844855.

Comment 12

[Julien Wajsberg [:julienw]]

So, not Aurora which is Gecko 21 now, but Gecko <= 20, so probably the b2g18 branch.

Comment 13

[John Ford [:jhford] CET/CEST Berlin Time]

v1-train@9bc1e1a

Comment 14

[Julien Wajsberg [:julienw]]

Forget my comments 11 and 12, I was misleaded. See more information in Bug 844855.

Comment 15

[Dan Mosedale (:dmosedale, :dmose)]

Any chance of getting this patch landed on v1.0.1 for the benefit/efficiency of people who working on that branch?

Comment 16

[Julien Wajsberg [:julienw]]

Just did it :

v1.0.1: 2a514d7f1d3f34512b9df8b538922f52f4f09832

Comment 17

[Dan Mosedale (:dmosedale, :dmose)]

Fantastic; thanks for the quick turnaround!