776785 - Alarm API - hal::SetAlarm takes |long|s, but should instead take PRInt32s

Status: RESOLVED FIXED

(Core :: Hardware Abstraction Layer (HAL), defect)

Description

[Justin Lebar (not reading bugmail)]

>+bool SetAlarm(long aSeconds, long aNanoseconds)

These should both be PRInt32's, as far as I can tell.  The only caller, AlarmHalService, has PRInt32s.

We avoid raw types in Gecko code because the meaning of |long| is different on different platforms.  (It's 64 bits wide on Linux 64, and 32 bits wide on Linux 32 and Windows 64.)

Is there a good reason that these are long's?  If not, we should fix it.

Comment 1

[Justin Lebar (not reading bugmail)]

Ping on this bug.  I can review the fix if you write the patch, Gene.

Comment 2

[Gene Lian (I already quit Mozilla)]

Thanks Justin! :) Sorry that I was busy with other issues. I'll come back with the patch today ASAP. Please stay tuned.

Comment 3

[Gene Lian (I already quit Mozilla)]

Attachment: Patch

Hi Justin,

Changes are trivial: replace the |long|s by |PRInt32|s. In addition to the hal:: functions, I also did the same thing for IDL for consistency. Please feel free to let me know if you didn't agree with these changes. Thanks for your reviews in advance :)

Comment 4

[Justin Lebar (not reading bugmail)]

Comment on attachment 647414 [details] [diff] [review]
Patch

Looks great; thanks!

Do you need me to check this in for you?

Comment 5

[Gene Lian (I already quit Mozilla)]

(In reply to Justin Lebar [:jlebar] from comment #4)
> Comment on attachment 647414 [details] [diff] [review]
> Patch
> 
> Looks great; thanks!
> 
> Do you need me to check this in for you?

Yes, thanks Justin! :) I don't have the privilege to check it in. Not yet pass the try sever (https://tbpl.mozilla.org/?tree=Try&rev=eb38577cd8d4) but I believe it works fine, since all the builds are successful and these changes won't affect the testing results. ;)

Comment 6

[Justin Lebar (not reading bugmail)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/bcf4a3c07234

Comment 7

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/bcf4a3c07234

Comment 8

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Note that we do use 'long' in IDL; the changes to IDL files in this bug were inappropriate.

Comment 9

[Justin Lebar (not reading bugmail)]

(In reply to :Ms2ger (Back and backlogged) from comment #8)
> Note that we do use 'long' in IDL; the changes to IDL files in this bug were
> inappropriate.

We also use PRInt32/PRUint32.  Is that deprecated for some bizarre reason?

$ git ls-files '*.idl' | xargs grep PRUint32 | wc -l
360
$ git ls-files '*.idl' | xargs grep PRInt32 | wc -l
102