1625461 - Build bustage: TypeError: unsupported operand type(s) for >>: 'builtin_function_or_method' and '_io.TextIOWrapper'.

Status: RESOLVED FIXED

(Toolkit :: Application Update, defect)

Description

[Ryan VanderMeulen [:RyanVM]]

Confirmed via local bisection that this was a regression from bug 1568994. If it matters, I build with the updater disabled.

build/application.ini.stub
build/application.ini.h.stub
toolkit/components/telemetry/TelemetryHistogramData.inc.stub
Traceback (most recent call last):
  File "C:\mozilla-build\python3\Lib\runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "C:\mozilla-build\python3\Lib\runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "c:\Users\Ryan\repos\mozilla\python\mozbuild\mozbuild\action\file_generate.py", line 124, in <module>
    sys.exit(main(sys.argv[1:]))
  File "c:\Users\Ryan\repos\mozilla\python\mozbuild\mozbuild\action\file_generate.py", line 72, in main
    ret = module.__dict__[method](output, *args.additional_arguments, **kwargs)
  File "c:/Users/Ryan/repos/mozilla/build/appini_header.py", line 35, in main
    print >>sys.stderr, \
TypeError: unsupported operand type(s) for >>: 'builtin_function_or_method' and '_io.TextIOWrapper'. Did you mean "print(<message>, file=<output_stream>)"?
mozmake.EXE[4]: *** [backend.mk;23: .deps/application.ini.h.stub] Error 1
mozmake.EXE[3]: *** [c:/Users/Ryan/repos/mozilla/config/recurse.mk;101: build/export] Error 2

Comment 1

[mac198442]

Yes. I have this patch to fix the prints in appini_header.py

diff --git a/build/appini_header.py b/build/appini_header.py
--- a/build/appini_header.py
+++ b/build/appini_header.py
@@ -27,18 +27,18 @@ def main(output, file):
                    for s in config.sections() for o in config.options(s))
     appdata['flags'] = ' | '.join(sorted(flags)) if flags else '0'
     appdata['App:profile'] = ('"%s"' % appdata['App:profile']
                               if 'App:profile' in appdata else 'NULL')
     expected = ('App:vendor', 'App:name', 'App:remotingname', 'App:version', 'App:buildid',
                 'App:id', 'Gecko:minversion', 'Gecko:maxversion', 'AppUpdate:url')
     missing = [var for var in expected if var not in appdata]
     if missing:
-        print >>sys.stderr, \
-            "Missing values in %s: %s" % (file, ', '.join(missing))
+        print("Missing values in %s: %s" % (file, ', '.join(missing)),
+              file=sys.stderr)
         sys.exit(1)
 
     if 'Crash Reporter:serverurl' not in appdata:
         appdata['Crash Reporter:serverurl'] = ''
 
     if 'App:sourcerepository' in appdata and 'App:sourcestamp' in appdata:
         appdata['App:sourceurl'] = '"%(App:sourcerepository)s/rev/%(App:sourcestamp)s"' % appdata
     else:
@@ -63,9 +63,10 @@ def main(output, file):
                  "%(AppUpdate:url)s"https://mail.yahoo.com/d/folders/1
              };''' % appdata)
 
 
 if __name__ == '__main__':
     if len(sys.argv) != 1:
         main(sys.stdout, sys.argv[1])
     else:
-        print >>sys.stderr, "Usage: %s /path/to/application.ini" % sys.argv[0]
+        print("Usage: %s /path/to/application.ini" % sys.argv[0],
+              file=sys.stderr)

With that installed it is obvious from the build that it fails because of disabling updater.

Comment 2

[mac198442]

hmm i guess i need to attach that rather than just adding

Comment 3

[mac198442]

Attachment: appini-header.patch

Comment 4

[Ian Neal]

Print statement fix over in bug 1625466

Comment 5

[mac198442]

Can we remove Windows from the summary since this issue is OS independent?

Comment 6

[Robin Steuber (they/them) [:bytesized]]

I can reproduce this if I disable building the updater. I think that the problem is that I conditionally included the update URL in application.ini, but didn't ifdef out the handling of the URL. It would be quite messy to ifdef out everything that propagates the url through appinfo, so I think instead I'll just unconditionally include the URL in application.ini.

Comment 7

[Robin Steuber (they/them) [:bytesized]]

Attachment: Bug 1625461 - Fix bug where conditionally including the update URL in application.ini breaks building without the updater r=glandium

Comment 8

[mac198442]

I was wondering if something like this might be better

[AppUpdate]
#if MOZ_UPDATER
URL=https://aus5.mozilla.org/update/6/%PRODUCT%/%VERSION%/%BUILD_ID%/%BUILD_TARGET%/%LOCALE%/%CHANNEL%/%OS_VERSION%/%SYSTEM_CAPABILITIES%/%DISTRIBUTION%/%DISTRIBUTION_VERSION%/update.xml
#else
URL=""
#endif

Comment 9

[Robin Steuber (they/them) [:bytesized]]

That seems reasonable. I'll update the patch.

Comment 10

[(No longer employed by Mozilla) Aaron Klotz]

Would you mind choosing a reviewer who is still working right now?

I ask because my (and probably others') local GeckoView build is broken right now without manually applying this patch, and assuming that glandium isn't coming back to work until Monday in Japan, you won't see any response until our Sunday night at the earliest.

Comment 11

[Robin Steuber (they/them) [:bytesized]]

Yeah. Let me find someone.

Comment 12

[[ex-Mozilla] Agi Sferro | :agi]

Any update? local build is busted for everybody right now :/

Comment 13

[Robin Steuber (they/them) [:bytesized]]

I've got a patch up that seems to fix things. Just waiting for code review.

Comment 14

[(No longer employed by Mozilla) Aaron Klotz]

Please find a reviewer for this.

Comment 15

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/91721122ded4
Fix bug where conditionally including the update URL in application.ini breaks building without the updater r=rstewart

Comment 16

[mac198442]

I don't get people complaining that this breaks local builds. exactly how hard is it to NOT specify --disable-updater in your local .mozconfig?

Comment 17

[Robin Steuber (they/them) [:bytesized]]

It breaks geckoview unconditionally, I assume because it never builds an updater.

Comment 18

[[ex-Mozilla] Agi Sferro | :agi]

Correct. GV doesn't use the updater. Also this broke lint automation too: https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=295662564&repo=try&lineNumber=1043

Comment 19

[mac198442]

well then the priority on this should be set to blocker.

Comment 20

[Natalia Csoregi [:nataliaCs]]

https://hg.mozilla.org/mozilla-central/rev/91721122ded4