Closed Bug 721496 Opened 13 years ago Closed 13 years ago

Remove code ifdef'd MOZ_WINSDK_TARGETVER for pre-Windows 7 SDKs

Categories

(Firefox Build System :: General, defect)

All
Windows 7
defect
Not set
minor

Tracking

(Not tracked)

RESOLVED FIXED
mozilla13

People

(Reporter: rain1, Assigned: capella)

References

()

Details

(Whiteboard: [good first bug])

Attachments

(1 file, 4 obsolete files)

Everything ifdef'd as MOZ_WINSDK_TARGETVER < MOZ_NTDDI_WIN7 should go.
Depends on: RequireWin7SDK
Whiteboard: [good first bug]
Severity: normal → minor
Summary: Remove code ifdef'd for pre-Windows 7 SDKs → Remove code ifdef'd MOZ_WINSDK_TARGETVER for pre-Windows 7 SDKs
Version: unspecified → Trunk
Depends on: 722370
"Found 142 matching lines in 41 files" Feel free to do this in multiple parts, maybe in blocking bugs too.
So looking at this request and without discussing it with anyone yet... With CONFIGURE.IN having: AC_DEFINE_UNQUOTED(MOZ_NTDDI_WS03, 0x05020000) AC_DEFINE_UNQUOTED(MOZ_NTDDI_LONGHORN, 0x06000000) AC_DEFINE_UNQUOTED(MOZ_NTDDI_WIN7, 0x06010000) Case #1 ===> (2) unused references to MOZ_NTDDI_WS03 seems safe to remove: /configure.in line 977 -- AC_DEFINE_UNQUOTED(MOZ_NTDDI_WS03, 0x05020000) /js/src/configure.in (View Hg log or Hg annotations) line 933 -- AC_DEFINE_UNQUOTED(MOZ_NTDDI_WS03, 0x05020000) Is Edge Case #2 ===> (1) reference to < MOZ_NTDDI_LONGHORN safe to change / consistent with this request?: /dom/plugins/ipc/PluginModuleChild.cpp line 692 -- #if !defined XP_WIN || MOZ_WINSDK_TARGETVER < MOZ_NTDDI_LONGHORN This code: ---------------------------- bool PluginModuleChild::RecvSetAudioSessionData(const nsID& aId, const nsString& aDisplayName, const nsString& aIconPath) { #if !defined XP_WIN || MOZ_WINSDK_TARGETVER < MOZ_NTDDI_LONGHORN NS_RUNTIMEABORT("Not Reached!"); return false; #else nsresult rv = mozilla::widget::RecvAudioSessionData(aId, aDisplayName, aIconPath); NS_ENSURE_SUCCESS(rv, true); // Bail early if this fails // Ignore failures here; we can't really do anything about them mozilla::widget::StartAudioSession(); return true; #endif } Would become This code: ---------------------------- bool PluginModuleChild::RecvSetAudioSessionData(const nsID& aId, const nsString& aDisplayName, const nsString& aIconPath) { #if defined XP_WIN && MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN nsresult rv = mozilla::widget::RecvAudioSessionData(aId, aDisplayName, aIconPath); NS_ENSURE_SUCCESS(rv, true); // Bail early if this fails // Ignore failures here; we can't really do anything about them mozilla::widget::StartAudioSession(); return true; #else NS_RUNTIMEABORT("Not Reached!"); return false; #endif }
(In reply to Mark Capella [:capella] from comment #3) > So looking at this request and without discussing it with anyone yet... > > With CONFIGURE.IN having: > > AC_DEFINE_UNQUOTED(MOZ_NTDDI_WS03, 0x05020000) > AC_DEFINE_UNQUOTED(MOZ_NTDDI_LONGHORN, 0x06000000) > AC_DEFINE_UNQUOTED(MOZ_NTDDI_WIN7, 0x06010000) > > Case #1 ===> (2) unused references to MOZ_NTDDI_WS03 seems safe to remove: If MOZ_NTDDI_WS03 isn't used anywhere, sure, it's safe to remove now. > > Is Edge Case #2 ===> (1) reference to < MOZ_NTDDI_LONGHORN safe to change / > consistent with this request?: > > /dom/plugins/ipc/PluginModuleChild.cpp > line 692 -- #if !defined XP_WIN || MOZ_WINSDK_TARGETVER < > MOZ_NTDDI_LONGHORN This should become #ifndef XP_WIN. > Would become This code: > ---------------------------- > bool > PluginModuleChild::RecvSetAudioSessionData(const nsID& aId, > const nsString& aDisplayName, > const nsString& aIconPath) > { > #if defined XP_WIN && MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN no, just #ifdef XP_WIN. We can now assume that MOZ_WINSDK_TARGETVER is always greater than MOZ_NTDDI_LONGHORN (and WIN7 for that matter).
Attached patch First Cut at changes (obsolete) — Splinter Review
I should have asked you to assign the bug to me, but I went ahead and started on it... This attached DIFF is code changes where I removed references to MOZ_NTDDI_WS03 and MOZ_NTDDI_LONGHORN ... (25) files .IN / .H / .CPP changed ... (I didn't remove references to MOZ_NTDDI_WIN7 yet...) I can go ahead and make the changes to remove MOZ_NTDDI_WIN7 references and re-build and retest if you want. I wasn't sure how a change of this size would be tested / moved forward (pieces or all at once) I ran a build and tested the browser, then ran the PYTEST suite and got 140,682 passes, 9 fails, and 3953 ToDo's. I have the .LOG file if you need to see it. Mozilla/5.0 (Windows NT 6.1; WOW64; rv:13.0a1) Gecko/20120214 Firefox/13.0a1 --mark
Attachment #596985 - Flags: feedback?(sagarwal)
Assignee: nobody → markcapella
Hardware: x86_64 → All
Status: NEW → ASSIGNED
Comment on attachment 596985 [details] [diff] [review] First Cut at changes > >-#if MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN > In cases where there is blank space both before and after the removed line, you should reduce the blank space too. >-#if MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN > // For Vista IFileDialog interfaces > #if _WIN32_WINNT < _WIN32_WINNT_LONGHORN > #define _WIN32_WINNT_bak _WIN32_WINNT > #undef _WIN32_WINNT > #define _WIN32_WINNT _WIN32_WINNT_LONGHORN > #define _WIN32_IE_bak _WIN32_IE > #undef _WIN32_IE > #define _WIN32_IE _WIN32_IE_IE70 > #endif >-#endif I'm not sure offhand but I think the Windows 7 SDK might default to newer values here too. >-#if defined(XP_WIN) && MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN > REGISTER(Private); >-#endif This change looks wrong.
Attached patch Second Cut at changes (obsolete) — Splinter Review
>>> In cases where there is blank space both before and after the removed line, you should reduce the blank space too. Ok, I Fixed up (5) files: windowsbattery.cpp removed (1) empty line nsdownloadserver.cpp removed (2) empty lines audiosession.cpp removed (2) empty lines audiosession.h removed (2) empty lines nsfilepicker.cpp removed (9) empty lines >>> I'm not sure offhand but I think the Windows 7 SDK might default to newer values here too. Fixed up file nsfilepicker.h by removing two blocks: - #if MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN - // For Vista IFileDialog interfaces - #if _WIN32_WINNT < _WIN32_WINNT_LONGHORN - #define _WIN32_WINNT_bak _WIN32_WINNT - #undef _WIN32_WINNT - #define _WIN32_WINNT _WIN32_WINNT_LONGHORN - #define _WIN32_IE_bak _WIN32_IE - #undef _WIN32_IE - #define _WIN32_IE _WIN32_IE_IE70 - #endif - #endif - -#if MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN -#if defined(_WIN32_WINNT_bak) -#undef _WIN32_WINNT -#define _WIN32_WINNT _WIN32_WINNT_bak -#undef _WIN32_IE -#define _WIN32_IE _WIN32_IE_bak -#endif -#endif - >>> This change looks wrong. Agrees :-) .......... fixed nsmemoryreportmanager.cpp as: #if defined(XP_WIN) REGISTER(Private); #endif
Attachment #596985 - Attachment is obsolete: true
Attachment #596985 - Flags: feedback?(sagarwal)
Attachment #597778 - Flags: feedback?(neil)
> >>> I'm not sure offhand but I think the Windows 7 SDK might default to newer values here too. > > Fixed up file nsfilepicker.h by removing two blocks: > > - #if MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN > - // For Vista IFileDialog interfaces > - #if _WIN32_WINNT < _WIN32_WINNT_LONGHORN > - #define _WIN32_WINNT_bak _WIN32_WINNT > - #undef _WIN32_WINNT > - #define _WIN32_WINNT _WIN32_WINNT_LONGHORN > - #define _WIN32_IE_bak _WIN32_IE > - #undef _WIN32_IE > - #define _WIN32_IE _WIN32_IE_IE70 > - #endif > - #endif > - > > -#if MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN > -#if defined(_WIN32_WINNT_bak) > -#undef _WIN32_WINNT > -#define _WIN32_WINNT _WIN32_WINNT_bak > -#undef _WIN32_IE > -#define _WIN32_IE _WIN32_IE_bak > -#endif > -#endif > - Try run that change - I don't think it will build. But it might!
Note, my mc mozilla-config.h has #define _WIN32_WINNT 0x502 http://mxr.mozilla.org/mozilla-central/source/configure.in#968
Built / running with it right now ....
(In reply to Jim Mathies from comment #9) > Note, my mc mozilla-config.h has > > #define _WIN32_WINNT 0x502 > > http://mxr.mozilla.org/mozilla-central/source/configure.in#968 Hmm, so should we be bumping WINVER too?
(In reply to neil@parkwaycc.co.uk from comment #11) > (In reply to Jim Mathies from comment #9) > > Note, my mc mozilla-config.h has > > > > #define _WIN32_WINNT 0x502 > > > > http://mxr.mozilla.org/mozilla-central/source/configure.in#968 > > Hmm, so should we be bumping WINVER too? Don't think so, this current targets server 2003 - This targets the Windows 2000 operating system. Other valid values include 0x0501 for Windows XP, 0x0502 for Windows Server 2003, 0x0600 for Windows Vista, and 0x0601 for Windows 7. http://msdn.microsoft.com/en-us/library/6sehtctf.aspx
(In reply to neil@parkwaycc.co.uk from comment #11) > Hmm, so should we be bumping WINVER too? No, WINVER is the oldest version of Windows code's guaranteed to run on. Seems like it was last bumped up (to Win2k3/XP SP2) by http://hg.mozilla.org/mozilla-central/rev/11c34494c1f3.
(In reply to Mark Capella [:capella] from comment #10) > Built / running with it right now .... Doesn't build locally for me - nsFilePicker.cpp f:\mozilla\firefox\mc\widget\windows\nsFilePicker.h(67) : error C2504: 'IFileDialogEvents' : base class undefined IFileDialogEvents are only defined if _WIN32_IE >= _WIN32_IE_IE60, and our current config has this set to 0x0500. _WIN32_IE_IE60 is defined as 0x0600 in sdkddkvers.h. Maybe we should stop defining this, as the sdkddkvers file sets if appropriately if it's not defined - #ifndef _WIN32_IE #ifdef _WIN32_WINNT // set _WIN32_IE based on _WIN32_WINNT #if (_WIN32_WINNT <= _WIN32_WINNT_NT4) #define _WIN32_IE _WIN32_IE_IE50 #elif (_WIN32_WINNT <= _WIN32_WINNT_WIN2K) #define _WIN32_IE _WIN32_IE_IE501 #elif (_WIN32_WINNT <= _WIN32_WINNT_WINXP) #define _WIN32_IE _WIN32_IE_IE60 #elif (_WIN32_WINNT <= _WIN32_WINNT_WS03) #define _WIN32_IE 0x0602 #else #define _WIN32_IE 0x0800 #endif #else #define _WIN32_IE 0x0800 #endif #endif
Yeah, IE6 or above seems safe to me!
doing a full clobber build now with these two lines removed - http://mxr.mozilla.org/mozilla-central/source/configure.in#969
So this works, although we will still have to up _WIN32_IE in places like nsFilePicker.h - w/o configure setting _WIN32_IE: _WIN32_IE=0x0602 NTDDI_VERSION=0x5020000 _WIN32_WINNT=0x502 after upping in nsFilePicker.h: _WIN32_IE=0x0700 NTDDI_VERSION=0x06000000 _WIN32_WINNT=0x0600 We need _WIN32_IE=_WIN32_IE_IE70 to get at those interfaces. Doesn't look like there's much else to change, we don't do this very often - http://mxr.mozilla.org/mozilla-central/search?string=_WIN32_IE Although there are some cases where we can get rid of _WIN32_WINNT upping code - http://mxr.mozilla.org/mozilla-central/search?string=_WIN32_WINNT
(In reply to Siddharth Agarwal from comment #13) > (In reply to comment #11) > > Hmm, so should we be bumping WINVER too? > > No, WINVER is the oldest version of Windows code's guaranteed to run on. > Seems like it was last bumped up (to Win2k3/XP SP2) by > http://hg.mozilla.org/mozilla-central/rev/11c34494c1f3. Odd, since that was 2009, and we only just dropped W2K...
Ok, dropping back and re-doing a clobber build did fail for me also this time. (Not sure what I missed). So then, are we asking me to correct by: Removing (2) lines from CONFIGURE.IN: - # Require OS features provided by IE 5.0 - AC_DEFINE_UNQUOTED(_WIN32_IE,0x0500) Allowing nsfilepicker to bump / drop VARS as in my first DIFF: -#if MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN // For Vista IFileDialog interfaces #if _WIN32_WINNT < _WIN32_WINNT_LONGHORN #define _WIN32_WINNT_bak _WIN32_WINNT #undef _WIN32_WINNT #define _WIN32_WINNT _WIN32_WINNT_LONGHORN #define _WIN32_IE_bak _WIN32_IE #undef _WIN32_IE #define _WIN32_IE _WIN32_IE_IE70 #endif -#endif -#if MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN #if defined(_WIN32_WINNT_bak) #undef _WIN32_WINNT #define _WIN32_WINNT _WIN32_WINNT_bak #undef _WIN32_IE #define _WIN32_IE _WIN32_IE_bak #endif -#endif Or did I miss something yet? I've got my clobber build running again, so I guess I'll find out ...
(In reply to Mark Capella [:capella] from comment #19) > Ok, dropping back and re-doing a clobber build did fail for me also this > time. (Not sure what I missed). > > So then, are we asking me to correct by: > > Removing (2) lines from CONFIGURE.IN: > > - # Require OS features provided by IE 5.0 > - AC_DEFINE_UNQUOTED(_WIN32_IE,0x0500) Hmm, since this is unrelated, I'm tempted to say lets file this in a separate bug and post this patch there. kheuy or ted should probably get an r? for this. > > Allowing nsfilepicker to bump / drop VARS as in my first DIFF: > > -#if MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN > // For Vista IFileDialog interfaces > #if _WIN32_WINNT < _WIN32_WINNT_LONGHORN > #define _WIN32_WINNT_bak _WIN32_WINNT > #undef _WIN32_WINNT > #define _WIN32_WINNT _WIN32_WINNT_LONGHORN > #define _WIN32_IE_bak _WIN32_IE > #undef _WIN32_IE > #define _WIN32_IE _WIN32_IE_IE70 > #endif > -#endif > > -#if MOZ_WINSDK_TARGETVER >= MOZ_NTDDI_LONGHORN > #if defined(_WIN32_WINNT_bak) > #undef _WIN32_WINNT > #define _WIN32_WINNT _WIN32_WINNT_bak > #undef _WIN32_IE > #define _WIN32_IE _WIN32_IE_bak > #endif > -#endif yep - that looks right.
Attached patch Third Try a charm... (obsolete) — Splinter Review
Changes as up through comment #20 ... Left configure.in alone as in original DIFF Left nsfilepicker.h alone as in original DIFF Asking :khuey for review, as I thought a saw a comment from :ted in IRQ about soon to be away for a while.
Attachment #597778 - Attachment is obsolete: true
Attachment #597778 - Flags: feedback?(neil)
Attachment #597904 - Flags: review?(khuey)
Note, you can scrap all changes in nsFilePicker from the patches here, I've landed those changes with bug 718374.
Attached patch Fourth Try (obsolete) — Splinter Review
Removed changes to nsfilepicker ...
Attachment #597904 - Attachment is obsolete: true
Attachment #597904 - Flags: review?(khuey)
Attachment #598533 - Flags: feedback?(jmathies)
Whichever patch it was I was requested on way back when, I did some testing with it, and I concluded that fiddling with _WINNT_WIN32 and/or _WINNT_IE wasn't going to allow the filepicker to compile, but the rest of the changes were OK.
Ok, I guess I'm waiting for a review+ or other help to do a push to the try server.
Comment on attachment 598533 [details] [diff] [review] Fourth Try (In reply to neil@parkwaycc.co.uk from comment #24) > Whichever patch it was I was requested on way back when, I did some testing > with it, and I concluded that fiddling with _WINNT_WIN32 and/or _WINNT_IE > wasn't going to allow the filepicker to compile, but the rest of the changes > were OK. So basically, r+ on the patch from you Neil?
Attachment #598533 - Flags: feedback?(jmathies) → review?(neil)
Comment on attachment 598533 [details] [diff] [review] Fourth Try >- AC_DEFINE_UNQUOTED(MOZ_NTDDI_WS03, 0x05020000) > AC_DEFINE_UNQUOTED(MOZ_NTDDI_LONGHORN, 0x06000000) Well, the other patch removed this define too, but r=me either way.
Attachment #598533 - Flags: review?(neil) → review+
hmmm ... thought I saw nsFilePicker::AppendFilter still references MOZ_NTDDI_LONGHORN after :jimm's patch and before mine/this one ...
(In reply to Mark Capella [:capella] from comment #29) > hmmm ... thought I saw nsFilePicker::AppendFilter still references > MOZ_NTDDI_LONGHORN after :jimm's patch and before mine/this one ... Ugh, you're right, I missed one. http://mxr.mozilla.org/mozilla-central/source/widget/windows/nsFilePicker.cpp#1246
I can fix the nsFilePicker.cpp and re-post a new DIFF ... (yes/do it?) I assume comment #27 was a push to try whose results I haven't seen yet ... (yes?) Should I be leaving MOZ_NTDDI_LONGHORN defined anyhow, as code search finds other: 73 matching lines in 24 files (yes?)
(In reply to Mark Capella [:capella] from comment #31) > I can fix the nsFilePicker.cpp and re-post a new DIFF ... (yes/do it?) yes please, sorry I missed that one. > I assume comment #27 was a push to try whose results I haven't seen yet > ... (yes?) yes - it'll post results here or you can view the results page. Builds look ok thus far. https://tbpl.mozilla.org/?noignore=0&tree=Try&rev=4570b9554147 > Should I be leaving MOZ_NTDDI_LONGHORN defined anyhow, as code search > finds other: 73 matching lines in 24 files (yes?) won't your patch remove all remaining references?
Try run for 4570b9554147 is complete. Detailed breakdown of the results available here: https://tbpl.mozilla.org/?tree=Try&rev=4570b9554147 Results (out of 47 total builds): exception: 1 success: 39 warnings: 3 failure: 4 Builds (or logs if builds failed) available at: http://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/jmathies@mozilla.com-4570b9554147
Attached patch Fifth PatchSplinter Review
Ok, I removed MOZ_NTDDI_LONGHORN from two configure.in files, and updated one missed item in nsFilePicker.cpp. I successfully ran a local build afterwards but am asking for another review? to be safe ... I can do an AUTOLAND to TRY after that to be thorough ... Dis-regard my comments about MOZ_NTDDI_LONGHORN. I was looking ahead to outstanding changes that might be left here to remove MOZ_NTDDI_WIN7 as per original comment, and comment # 4, and my brain got full -- mark
Attachment #598533 - Attachment is obsolete: true
Attachment #598903 - Flags: review?(jmathies)
Attachment #598903 - Flags: review?(jmathies) → review+
Keywords: checkin-needed
Status: ASSIGNED → RESOLVED
Closed: 13 years ago
Resolution: --- → FIXED
Depends on: 730128
Product: Core → Firefox Build System
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: