What’s wrong with Visual C++ and Microsoft Connect?

Using Visual Stu­dio 2010 with Ser­vice Pack 1, try doing the following:

1. cre­ate a new project of type Win32 Console Application, and under Application Settings in the project wiz­ard, select Console Application and Empty Project.
2. cre­ate a sin­gle .cpp file
3. add the code int main() {} to it
4. hit build.

You’ll now get the warn­ing (listed under “Errors”, for some rea­son): IntelliSense: PCH warning: header stop cannot be in a macro or #if block. An intellisense PCH file was not generated.

(I zipped up a solu­tion repro­duc­ing the prob­lem here. Note that every­thing is left at the default settings.

What the f…? How can some­thing so sim­ple break IntelliSense?

I am some­times amazed at the appar­ent lack of test­ing that goes into a release of Visual C++. VC2010 beta 1 had a glar­ingly obvi­ous bug, where the above pro­gram would fail with a linker error. Appar­ently, no one had thought to test a pro­gram which used the standard-compliant main instead of wmain with UNICODE enabled. Sure, it was beta, and you’d expect errors, but some­thing sim­pler than a Hello World should still work, shouldn’t it?

Now we get spu­ri­ous Intel­liSense errors if we have the audac­ity to cre­ate the tini­est, sim­plest pro­gram imag­in­able, at the default settings.

Of course, it gets bet­ter. Because as we all know, Microsoft Con­nect is beyond use­less, and obvi­ously, some optimistic/foolish VC++ users had already reported the prob­lem with this warn­ing mes­sage. Google gave me the fol­low­ing three bug reports:

• this one, closed as By Design
• this, closed as Dupli­cate
• And of course, no dis­cus­sion of Con­nect would be com­plete with­out some­thing closed as Won’t Fix

It is one thing that errors that prac­ti­cally leap in the user’s face are allowed to exist in a ship­ping prod­uct at all, that lit­er­ally the short­est, sim­plest pro­gram pos­si­ble pro­vokes the bug, but that Microsoft closes such bugs with three dif­fer­ent close rea­sons only serves to under­line how utterly use­less, ran­dom and user–hos­tile Microsoft Con­nect is, and it is pretty hard to escape the con­clu­sion that there is some­thing fun­da­men­tally wrong with how Microsoft’s Dev­Div operates.

I am gen­er­ally rea­son­ably pos­i­tive towards Microsoft’s C++ com­piler. It’s noth­ing spe­cial, cer­tainly, and Microsoft’s pol­icy of never ever improv­ing an exist­ing prod­uct means it is per­pet­u­ally obso­lete com­pared to the com­pe­ti­tion, but it gen­er­ally works, and it has made great strides towards stan­dards com­pli­ance. But the qual­ity of the IDE, and of Intel­lisense, is really absurd, at times. VC2010 seems to have shifted the absur­dity of Intel­liSense from “it just plain doesn’t work”, to “It works rea­son­ably ok, but imposes some baf­fling require­ments on the code, has some kind of love affair with pre­com­piled head­ers and seem­ingly doesn’t even bother with code that doesn’t have one, reports false pos­i­tive ‘errors’, and lists them in the ‘Errors’ win­dow, along with actual com­pile errors.”

And Con­nect is noth­ing short of an insult towards Microsoft’s cus­tomers. It is appar­ently staffed by mon­keys, per­haps assisted by a script that ran­domly closes issues as “Won’t Fix”, and seems to serve the sole pur­pose of under­lin­ing to Microsoft’s cus­tomers, all the errors that will not get fixed.

Here’s another great exam­ple I saw in my Twit­ter feed just two days ago:

#SystemIsBroke indeed. Why even have Con­nect, when bugs appar­ently only get resolved once some­one with lots of fol­low­ers on Twit­ter com­plain about it?

And of course, this gem of a response from Microsoft on the bug report itself is really remarkable:

I do under­stand your con­cern, but we are beyond our ZBB date, there­fore only ship-blocker bugs make the bar at this point. Con­sid­er­ing that we have shipped this bug before, it is hard to make a case that we need to fix it in this release since it did not block our pre­vi­ous release. Thank you.

Wait what? I guess this pretty much con­firms what we all knew: that once a bug has been closed as “Won’t Fix”, the claims that “we will eval­u­ate this for a future release” are just pre­tense. What actu­ally hap­pens is that the older a bug gets, the lower pri­or­ity it’s given, accord­ing to some twisted “prece­dence” prin­ci­ple. “We’ve been able to ignore the bug for X years now, so surely, there’d be no harm in ignor­ing it one more year”.

Now sure, I’m not par­tic­u­larly con­cerned about WPF, but it shows that the prob­lems with Con­nect are not lim­ited to Visual C++ issues.

Dear Microsoft, Con­nect is bro­ken. And every­one except you knows it.

Tags: c++, connect, rants, visual-studio

Programming | RSS 2.0 | Respond | Trackback |