Chromium Being Ported To VC++, Scrubbed of Compiler Bugs

← Back to Stories (view on slashdot.org)

Chromium Being Ported To VC++, Scrubbed of Compiler Bugs

Posted by timothy on Friday March 25, 2016 @03:01AM from the sounds-like-a-voice-warm-up dept.

jones_supa writes: Moving a big software project to a new compiler can be a lot of work, and few projects are bigger than the Chromium web browser. In addition to the main Chromium repository, which includes all of WebKit, there are over a hundred other open-source projects which Chromium incorporates by reference, totaling more than 48,000 C/C++ files and 40,000 header files. As of March 11th, Chromium has switched to Visual C++ 2015, and it doesn't look like it's looking back. The tracking bug for this effort currently has over 330 comments on it, with contributions from dozens of developers. Bruce Dawson has written an interesting showcase of some VC++ compiler bugs that the process has uncovered. His job was to investigate them, come up with a minimal reproduce case, and report them to Microsoft. The Google and Microsoft teams get praise for an excellent symbiotic relationship, and the compiler bugs have been fixed quickly by the Visual Studio team.

59 of 93 comments (clear)

Min score:

Reason:

Sort:

Re:And all so they can sell you to advertisers... by Kkloe · 2016-03-25 03:09 · Score: 1

like internet, smartphones and computers?, do you REALLY REALLY need thoose?
Why? by guruevi · 2016-03-25 03:17 · Score: 1, Insightful

Most developers know VC++ compilers are full of bugs and weird stuff. Why didn't they just stay with the compilers that are well supported across all platforms?

--
Custom electronics and digital signage for your business: www.evcircuits.com
1. Re:Why? by Anonymous Coward · 2016-03-25 03:36 · Score: 4, Interesting
  
  All compilers have 'weird bugs and stuff' - staying with compilers 'that are well support across all platforms' is another matter altogether. I'm not sure why you are asserting the fact that VC++ has bugs - your text reads as if you are trying to convey it's the only compiler that does contain bugs, therefore you can't fathom as to why on earth a development team would move from one dev platform to another because of the bugs. Because it's Microsoft, it must be shit right? That's what you're saying?
2. Re:Why? by guruevi · 2016-03-25 04:12 · Score: 1, Insightful
  
  I'm not saying they are the only compiler with bugs, it's just that VC++ is known (at least in academia that write shitty software to begin with) to generate a lot more errors/bugs for software that's not directly Win32 related than eg. GCC or Clang. Not sure what the rationale was behind going to a closed software compiler with possible patent issues/assertions.
  
  --
  Custom electronics and digital signage for your business: www.evcircuits.com
3. Re:Why? by Anonymous Coward · 2016-03-25 04:44 · Score: 5, Insightful
  
  A lot of times compiler bugs fall into categories that aren't bugs at all. 1. Stuff that people think is standard C, but it's really a GCC-ism. 2. Stuff people didn't know about standard C, such as the order of argument evaluation being unspecified. 3. Default settings on one compiler being forgiving and/or handling uninitialized variables in a particular way, and the other compiler not being so forgiving.
  I can't remember who said so first, but it's generally given that the first people to scream "compiler bug" are usually doing something wrong with their code. I wouldn't be surprised if a lot of the people in academia who you say write shitty software are doing that kind of thing. "Oh! compiler bug!" Nope. You just had something that "worked on my machine" and now it doesn't.
4. Re:Why? by jellomizer · 2016-03-25 05:19 · Score: 1
  
  I expect it is a point of optimization.
  While VC++ may have a bunch of bugs, it may be due to over optimization, for faster performance on Windows Platforms.
  
  --
  If something is so important that you feel the need to post it on the internet... It probably isn't that important.
5. Re:Why? by TheRaven64 · 2016-03-25 06:02 · Score: 2
  
  Google people have been investing a lot of time and effort over the last year or so in Windows support for clang. Interestingly, Microsoft has also been contributing and has some odd combinations (MSVC++ front end with LLVM back end and Clang front end with MSVC++ back end) working.
  
  --
  I am TheRaven on Soylent News
6. Re:Why? by armanox · 2016-03-25 06:29 · Score: 2
  
  I don't know, GCC has a lot of quirks that make it a pretty poor standard to code by. By making it compile with VC++ they open the door to elminiting the GCC workarounds, and possibly open the door for more platforms. As someone who has tried to port software in the past, GCC is really a thorn in my side.
  
  --
  I'm starting to think GNU is the problem with "GNU/Linux" these days.
7. Re:Why? by Anonymous Coward · 2016-03-25 06:30 · Score: 2, Informative
  
  I have ported around thru a bunch of different compilers over the years a few codebases.
  I can count on one hand the number of VC compiler bugs I have come across. They are seriously nasty to root out as my assumption is the compiler is usually right. MS usually had it fixed in under a week with a hotfix. When you buy visual studio you get some fairly top notch support.
  Now the other bugs. I lost count of. All of those were usually a misunderstanding of how C/C++ works. A few cases were like you said GCC-isms vs VC-isms. Where the code would do one thing on one compiler vs the other. Many times it was we were doing something that was not quite in spec anyway and it was just easier to toss out the silly code and re-write that small bit.
  Also most of the time the big thing to keep in mind is the standard libraries are not the same across platforms. Take little old printf and scanf. That is C 150 sort of stuff you learn. Yet that thing is a land mine in pretty much all compiler suites. Then it acts differently on all of them (things like does %hhu work with this one or not, or it works with printf but not scanf sort of junk). Those two functions are a first class ticket to buffer overruns and underruns. They are all getting better on agreeing with things should do. But it is taking time.
  Another fun gotcha between compilers is padding. Some do 16 bytes some do 8. Not a big deal right? Right up until the point you want to tear it apart and give that info to some other process. That may or may not be written with the same padding. So dense pack right? That ends up in memory alignment hell. x86 is fairly forgiving on that. ARM not so much.
  Usually if I get it down to one compiler does X and the other does Y I abstract it. I abstract that small bit into its own file and then just switch out the files at build time or sometimes a #define but try not to do that can be confusing (whichever way is more clear I do).
  Also most optimizer compiler bugs come down to you passing in the same memory space as two different parameters into a function. Dont do that.
8. Re:Why? by Megol · 2016-03-25 06:39 · Score: 1
  
  Microsofts compilers have always been conservative when it comes to optimization. And frankly in most cases a smart compiler using the best optimization possible* will be at most a few percentages faster than the one that only does basic optimizations (like register allocation, common subexpression elimination etc.).
  GCC and Clang/LLVM have switched to the idea that all undefined types of code shouldn't produce what is likely to be the programmers intent like C compilers of old, nor should they flag undefined cases as errors and stop compiling (as that would break a huge amount of software) - instead it is an invitation to optimize like if the undefined case couldn't be. Sure it breaks software but unless you are important (read: SPEC benchmarks) it's your own damn fault. If you are important you'll be special cased instead in order to not break the code...
  (* this is a theoretical compiler - it doesn't exist)
9. Re:Why? by Billly+Gates · 2016-03-25 08:54 · Score: 1
  
  If an opensource developer must target Windows he or she can use clang with VS 2015 community edition just fine or vim?
  Still it is good to use a standards compliant api and c++ that is not tied too heavy on one product.
  In other news
  But the big elephant in the room is the switch to VC2015 means the death of XP and Vista.
  Is Chrome dropping support for XP finally? Will these users get ransomware from using an unpatched and unsupported browser now if Chromium (the foundation of Chrome) is not compatible?
  
  --
  http://saveie6.com/
10. Re:Why? by Anonymous Coward · 2016-03-25 11:28 · Score: 1
  
  >> Also most optimizer compiler bugs come down to you passing in the same memory space as two different parameters into a function. Dont do that.
  This is called memory aliasing. And compilers are usually pessimistic about optimizing when memory references can't be identified to be of each other.
  Just do not use "restrict" keyword if you are not sure. https://en.wikipedia.org/wiki/Restrict
11. Re:Why? by tlhIngan · 2016-03-25 20:11 · Score: 2
  
  I don't know, GCC has a lot of quirks that make it a pretty poor standard to code by. By making it compile with VC++ they open the door to elminiting the GCC workarounds, and possibly open the door for more platforms. As someone who has tried to port software in the past, GCC is really a thorn in my side.
  Another benefit is using multiple compilers helps find tricky bugs due to assumptions in the programming language.
  Enabling all warnings can easily help find those parts of code that are iffy but look normal, but the compiler knows something may be an issue. Using a different compiler produces a different set of messages that can reveal all sorts of bugs.
  
  But the big elephant in the room is the switch to VC2015 means the death of XP and Vista.
  Is Chrome dropping support for XP finally? Will these users get ransomware from using an unpatched and unsupported browser now if Chromium (the foundation of Chrome) is not compatible?
  Well, Chrome has been warning XP users that support for XP is stopping "shortly". The stupid thing is, the "Learn more" link the pop up shows leads to the "Download Chrome" page instead of a page explaining why. So yes, they're stopping support for XP and VIsta, but they aren't telling you why, other than to "Download Chrome".
12. Re:Why? by arglebargle_xiv · 2016-03-26 19:43 · Score: 1
  
  Most developers know VC++ compilers are full of bugs and weird stuff. Why didn't they just stay with the compilers that are well supported across all platforms?
  I do cross-platform development for code that has to run on everything (literally). It's plain C, not C++, so I'm saved from seeing various bugs by that. The VC++ compilers are some of the least buggy compilers I've ever used. I've actually never found a compiler by in VC++ 6.0 in nearly 20 years of use. In contrast, gcc is one of the buggiest (mainstream) compilers I've used, every new release seems to contain a new collection of code-generation bugs.
  In terms of more obscure compilers, even gcc pales in comparison to some proprietary ones, the worst compiler I've ever used was the SCO one, with the PHUX compiler coming a close second. Some of the Solaris compilers were pretty flaky as well. And Arm's compiler also gets pretty hairy once you start using some of the more exotic features some of the Arm architectures support, you really want to get that with a maintenance contract to make sure you've got all the latest patches.
Re:And all so they can sell you to advertisers... by BarbaraHudson · 2016-03-25 03:23 · Score: 1

More like smart watches, 3d tvs and vr headsets, do you really, really need those?

--
"Transparent" is a shit show that trades on every stereotype going. A man in drag is NOT a transsexual.
Windows @ Google by Luthair · 2016-03-25 03:25 · Score: 1

I thought this wasn't allowed and they had to choose Linux or OSX
Chrome + Windows = Mirror by mrthoughtful · 2016-03-25 03:26 · Score: 4, Funny

Can't see anything but your reflection..

--
This comment was written with the intention to opt out of advertising.
1. Re:Chrome + Windows = Mirror by Virtucon · 2016-03-25 03:48 · Score: 1
  
  that's actually silver + windows = mirror. #FTFY
  
  --
  Harrison's Postulate - "For every action there is an equal and opposite criticism"
2. Re:Chrome + Windows = Mirror by Livius · 2016-03-25 04:27 · Score: 2
  
  Chrome + Windows = suboptimal mirror.
3. Re:Chrome + Windows = Mirror by Solandri · 2016-03-25 10:53 · Score: 1
  
  I always wondered when someone would use the webcam to create a "chrome" desktop icon which reflected like real chrome.
Re:From what to VC? by Anonymous Coward · 2016-03-25 03:29 · Score: 4, Informative

As near as I can tell they switched the Windows build from VC++ old version to VC++ new version. Because:
- Improved C++11/14 support
- Security improvements
- Some /analyze checked rolled into main compile
- Possibly some improved support for "cloud" builds
Re:Is there a reason? by ausekilis · 2016-03-25 03:31 · Score: 1

This blog post doesn't offer any solid answers, though some of the comments point to this being a transition from VC2013 for the windows build. There's also a reddit post of the build/link taking near an hour for a 24 core machine.
Re:Is there a reason? by ausekilis · 2016-03-25 03:32 · Score: 1

... and then I realized my google search turned up one of the links in TFS. Is it the weekend yet?
That's huge! by QuietLagoon · 2016-03-25 03:33 · Score: 2

...there are over a hundred other open-source projects which Chromium incorporates by reference, totaling more than 48,000 C/C++ files and 40,000 header files....
That's a lot of files for a single, relatively trivial application..
1. Re:That's huge! by Carewolf · 2016-03-25 04:16 · Score: 4, Insightful
  
  Chrome is not an application. It is an entire operating system, that just happens to include a browser. Think of it as ChromeOS VM.
2. Re:That's huge! by Anonymous Coward · 2016-03-25 04:19 · Score: 1
  
  There's nothing "relatively trivial" about the modern web browser. A modern JIT optimizing javascript engine and basic class libraries alone make it as complicated as many compilers. DOM implementation + layout + rendering is also hugely complex. Quick and efficient local caching, plus history recall requires the complexity of a mid-range database server.
  There are many more parts I haven't even touched on.
3. Re:That's huge! by Dragonslicer · 2016-03-25 04:44 · Score: 4, Funny
  
  Chrome is not an application. It is an entire operating system, that just happens to include a browser.
  How does it compare to emacs?
4. Re:That's huge! by networkBoy · 2016-03-25 04:49 · Score: 1
  
  /hat tip
  
  --
  whois gawk date unzip strip find touch finger mount join nice man top fsck grep eject more yes exit umount sleep dump
5. Re:That's huge! by mrchaotica · 2016-03-25 04:57 · Score: 2
  
  It's GUI-based instead of ncurses-based, and is scripted with Javascript instead of Lisp.
  
  --
  "[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz
6. Re:That's huge! by Dragonslicer · 2016-03-25 05:08 · Score: 4, Funny
  
  It's GUI-based instead of ncurses-based
  Moderate advantage to Chrome on that one, I think.
  
  and is scripted with Javascript instead of Lisp.
  Err, we'll call that one a draw.
7. Re:That's huge! by QuietLagoon · 2016-03-25 05:08 · Score: 1
  
  Chrome is not an application.
  https://www.chromium.org/Home
  .
  "...Chromium is an open-source browser project ..."
8. Re:That's huge! by Carewolf · 2016-03-25 05:49 · Score: 1
  
  Chrome is not an application.
  https://www.chromium.org/Home
  .
  "...Chromium is an open-source browser project ..."
  Which is why I weighted my words and went with Chrome which is mainly a branded binary from the Chromium project. Where "Chromium" can be either the application or the project.
9. Re:That's huge! by Darinbob · 2016-03-25 09:39 · Score: 1
  
  Emacs is GUI based, it's only ncurses based if you're using it from a terminal. For GUI is it not tied to one particular type of GUI either.
  Lisp is a real language with lots of history and theory, JavaScript is a hack, it's not a draw it's a clear advantage to Emacs there.
10. Re:That's huge! by Anonymous Coward · 2016-03-25 10:43 · Score: 1
  
  Emacs Lisp is sort of a special beast. It only looks like Lisp to people that don't program in other Lisp varieties.
11. Re:That's huge! by fnj · 2016-03-25 11:17 · Score: 1
  
  Emacs is GUI based, it's only ncurses based if you're using it from a terminal.
  Or if you are using the one built without X support. I wouldn't have that bloat on my system for a fucking TEXT EDITOR. Not even on a system which has X; let alone a server with no X.
12. Re:That's huge! by Darinbob · 2016-03-25 12:23 · Score: 1
  
  I really like my Aquamacs though. Also having multiple open windows is nice. Using it only from a terminal would feel clumsy, even though that's what I used to do.
Re:From what to VC? by OzPeter · 2016-03-25 03:46 · Score: 2

Chromium has switched to Visual C++ 2015
What, from Borland Turbo C++ 3.0?
From thesecond link it seems as if they were upgrading from VC++ 2013

--
I am Slashdot. Are you Slashdot as well?
Re:Odd choice of tools by jszpilewski · 2016-03-25 03:54 · Score: 5, Informative

Just their principal Windows build environment changed in March 2016 from VS 2013 (still supported) to VS 2015. This is no news... really.
Re:Odd choice of tools by dbrueck · 2016-03-25 03:56 · Score: 2

The summary, especially the title, omitted a really important detail: it's not that they are porting it to VC++, it's that they are making it work with VC++ *2015*.
IOW they apparently already had the build working with an earlier version of VC++, and this is all about them changing things to work with this newer version of VC++.
Re:Odd choice of tools by gnupun · 2016-03-25 04:01 · Score: 1

Isn't Visual Studio free for open source projects?
It sounds like the compiler is looking OK now. by pentagramrex · 2016-03-25 04:07 · Score: 1

It used to be full of bugs, but they bought some of the best people. Working on Webkit is a pain in the arse. Huge compile times, and no documentation worth talking about.
Who cares about VC++? by LinuxFreakus · 2016-03-25 04:08 · Score: 1

How about they fix all the crappy unaligned casts all over the place so I can actually compile the source on alternative architectures and have it work (among other issues with that code that pose difficulties for cross platform work).
1. Re:Who cares about VC++? by pentagramrex · 2016-03-25 04:11 · Score: 1
  
  The compiler only targets x86, used to do arm too, look at embedded compilers if you want to see some horrible bugs.
2. Re:Who cares about VC++? by Cley+Faye · 2016-03-25 04:13 · Score: 1
  
  > they
  You're free to provide these fixes yourself if you care.
3. Re:Who cares about VC++? by LinuxFreakus · 2016-03-25 04:21 · Score: 1
  
  Yes, it does... that was exactly my critique... I wish they hadn't gotten lazy and only targeted x86 because it works pretty well.
4. Re:Who cares about VC++? by LinuxFreakus · 2016-03-25 04:26 · Score: 1
  
  Well, I could you're right. But it would be a lot of work because the code is littered with issues (unless you only care about x86). But there is also FireFox, which compiles and runs quite well on a variety of architectures with minimal patching aside from a few optional dependencies that I just leave out. I don't have a strong enough need for it to DIY (and it could be a fair point if people say that just proves that they don't need to target anything else, but IMO if they had just done it from the start it wouldn't have been much extra work... but now it would be pretty hard to fix it unless there were a compelling reason).
5. Re:Who cares about VC++? by religionofpeas · 2016-03-25 04:38 · Score: 1
  
  Just fix your crappy hardware to support unaligned access.
6. Re:Who cares about VC++? by DaHat · 2016-03-25 05:50 · Score: 1
  
  No it doesn't. At last check it supports building code which targets x86, x64, ARM and ARM64.
  
  --
  Help Brendan pay off his student loans
Slow news day or what by Anonymous Coward · 2016-03-25 04:26 · Score: 1

So Chromium is switching from an older version of VC-- to a newer version and this is considered news for nerds? WTF /.?
Impossible by Anonymous Coward · 2016-03-25 05:24 · Score: 1

Are you implying that the submission was flat out wrong? That's impossible.
Chromium Development by Tenebrousedge · 2016-03-25 07:29 · Score: 1

I don't know if you've looked at what it takes to set up a developer environment for Chromium, but that, to me, seems like a hell of a lot of time investment. It might be worth it if you really wanted something done, but for me that's crossing the line into "I want to be paid for that kind of work," and verging on, "You couldn't pay me to work on that." I don't have a lot of comparative experience with complex build environments, and it does seem to be well documented, but still...it seems quite the ordeal.

--
Those who advocate genocide deserve every protection afforded by law, and none afforded by common human decency.
1. Re:Chromium Development by Cley+Faye · 2016-03-25 09:06 · Score: 1
  
  I'm not going to look, but simply knowing that the thing work on multiple systems and have a gazillion of dependencies, I'm certain it is a nightmare to set up.
  But that's the point; if someone who need these big changes to happen won't go through it (and I completely understand why, it's not a reproach), asking for the people that didn't need these changes to work on them is kinda rude.
2. Re:Chromium Development by Anonymuous+Coward · 2016-03-25 09:46 · Score: 1
  
  quoting from that link:
  
  an account with sudo access You need root access to run the chroot command and to modify the mount table. NOTE: Do not run any of the commands listed in this document as root â" the commands themselves will run sudo to get root access when needed.
  
  There should be some harsh physical punishment for the idiots that put sudo calls in Makefiles. And for sudo-heads in general. Why would you need root just to build some fucking binary, should it be a hello-world or your browser/os of the day? Even for building a kernel for the machine it's built on, you need root only for the final step, where you install the kernel/modules/ramdisk.
  That piece of shit (sudo) should be banned once for all. If you feel like fucking up your systems and ignoring all the good security practices, go install your own setuid garbage; it's easy.
Re:And all so they can sell you to advertisers... by Anonymous Coward · 2016-03-25 07:32 · Score: 5, Informative

>"As of March 11th, Chromium has switched to Visual C++ 2015..."
This should have been written as:
"As of March 11th, WINDOWS Chromium build has switched supported compilers FROM Visual C++ 2013 TO Visual C++ 2015 ..."
This has nothing to do with OSX or linux builds and is not that much important news.
Re: And all so they can sell you to advertisers... by markhb · 2016-03-25 07:59 · Score: 1

Those of us so old we have 5-digit ID's, that's who! Now Get Off My Lawn!

--
Save Maine's economy: write stuff down. All comments are exclusively my own, not my employer.
Mozilla guys doing the same thing by Ilgaz · 2016-03-25 08:04 · Score: 1

Mozilla Firefox has a "bug" open for same purpose and there are several reasons for the switch, build performance increase and increased security.
https://bugzilla.mozilla.org/s...
Re:From what to VC? by Darinbob · 2016-03-25 09:16 · Score: 2

So Microsoft tools are so difficult to use across different tool versions that it becomes a news story when someone manages to upgrade?
Re: And all so they can sell you to advertisers... by narrowhouse · 2016-03-25 10:02 · Score: 2

Just because we let you cut the lawn doesn't make it yours sonny. :)

--

Insert pithy comment here.
Re:And all so they can sell you to advertisers... by Bruce+Dawson · 2016-03-28 11:55 · Score: 1

Fair enough. I didn't realize how ambiguous that sentence would appear, especially when the slashdot summary omitted '2015'. I fixed the sentence in my blog post.
> and is not that much important news
I thought it was interesting, which was why I wrote it. Whether it is important is up to each reader to decide.