Comments for Verilab Blog http://www.verilab.com/blog Verilab Thu, 25 Sep 2008 23:38:39 +0000 http://wordpress.org/?v=2.6.1 Comment on DAC 2008 Presentations Now Posted by JL Gray http://www.verilab.com/blog/2008/07/dac-2008-presentations-now-posted/#comment-163 JL Gray Thu, 25 Sep 2008 23:33:00 +0000 http://www.verilab.com/blog/2008/07/dac-2008-presentations-now-posted/#comment-163 Sean, I believe this should be fixed (again). We've made some other modifications which I hope will resolve the problem once and for all. If you're still seeing issues please let me know. Thanks, JL Sean,

I believe this should be fixed (again). We’ve made some other modifications which I hope will resolve the problem once and for all. If you’re still seeing issues please let me know.

Thanks,

JL

]]> Comment on DAC 2008 Presentations Now Posted by Sean http://www.verilab.com/blog/2008/07/dac-2008-presentations-now-posted/#comment-162 Sean Mon, 22 Sep 2008 02:17:09 +0000 http://www.verilab.com/blog/2008/07/dac-2008-presentations-now-posted/#comment-162 The posts feed is still compromised. (Or has been compromised again.) Drop me an email if you want so see the feed contents that I see. http://www.verilab.com/blog/feed/ on NewsGator. -Sean The posts feed is still compromised. (Or has been compromised again.)

Drop me an email if you want so see the feed contents that I see.

http://www.verilab.com/blog/feed/ on NewsGator.

-Sean

]]> Comment on DAC 2008 Presentations Now Posted by JL Gray http://www.verilab.com/blog/2008/07/dac-2008-presentations-now-posted/#comment-160 JL Gray Wed, 20 Aug 2008 15:49:48 +0000 http://www.verilab.com/blog/2008/07/dac-2008-presentations-now-posted/#comment-160 James, Thanks for the heads up on this. Turns out our blog software was somewhat out of date and a few of the posts/comments had been compromised. The issue should be resolved now. Let me know if you're still seeing any issues. Take care, JL James,

Thanks for the heads up on this. Turns out our blog software was somewhat out of date and a few of the posts/comments had been compromised. The issue should be resolved now. Let me know if you’re still seeing any issues.

Take care,

JL

]]> Comment on DAC 2008 Presentations Now Posted by James Colgan http://www.verilab.com/blog/2008/07/dac-2008-presentations-now-posted/#comment-159 James Colgan Sun, 17 Aug 2008 16:51:33 +0000 http://www.verilab.com/blog/2008/07/dac-2008-presentations-now-posted/#comment-159 Hi, We've implemented a feed reader on the Xuropa Platform and have found that the verilab RSS feed has been compromised. From the code, it looks like it's within the comments feed (a usual point of attack for spammers), even though there is a captcha system in place on your site. The junk may have been injected over at feedburner instead, but I can't tell from the code. We still have your feed on the platform, but the comments are a mess and won't be included. Also, it does make your feed very "heavy". I've emailed you the feed file separately. James Hi,

We’ve implemented a feed reader on the Xuropa Platform and have found that the verilab RSS feed has been compromised.

From the code, it looks like it’s within the comments feed (a usual point of attack for spammers), even though there is a captcha system in place on your site. The junk may have been injected over at feedburner instead, but I can’t tell from the code.

We still have your feed on the platform, but the comments are a mess and won’t be included. Also, it does make your feed very “heavy”.

I’ve emailed you the feed file separately.

James

]]> Comment on SystemVerilog Gotcha: (when copying) a struct is not a class by another name by Jason Sprott http://www.verilab.com/blog/2008/01/gotcha-when-copying-a-struct-is-not-a-class-by-a-different-name/#comment-155 Jason Sprott Wed, 23 Jan 2008 08:28:30 +0000 http://www.verilab.com/blog/2008/01/gotcha-when-copying-a-struct-is-not-a-class-by-a-different-name/#comment-155 Hi Daniel Thanks for pointing out that error (now corrected). I fixed the variable names and also added the new() of the elements in Y[], inside the foreach loop, for clarity. Yes, when a struct is copied the values are copied. When a class is copied the reference is copied, or a shallow copy is taken (as described in IEEE1800-2005, Section 7.11). I find people typically remember that when they are dealing with the class itself. When classes are inside something else, it's easy to forget, because we are copying the thing and may not even have visibility of its contents. Arrays of classes fall into that category, but a class composed of other classes is another big one to watch. There is no method for a deep copy/clone defined in the SystemVerilog language. Both the AVM and VMM define methods for copying objects as part of the methodology. In Vera we had a built-in method called object_copy() as part of the language. This provided a deep copy of any object, without you having to write anything yourself. I found it useful, but the problem was that there was no control, so you could end up copying something *very* large. Of course when writing custom copy functions we sometimes put bugs in them too. Jason Hi Daniel

Thanks for pointing out that error (now corrected). I fixed the variable names and also added the new() of the elements in Y[], inside the foreach loop, for clarity.

Yes, when a struct is copied the values are copied. When a class is copied the reference is copied, or a shallow copy is taken (as described in IEEE1800-2005, Section 7.11). I find people typically remember that when they are dealing with the class itself. When classes are inside something else, it’s easy to forget, because we are copying the thing and may not even have visibility of its contents. Arrays of classes fall into that category, but a class composed of other classes is another big one to watch.

There is no method for a deep copy/clone defined in the SystemVerilog language. Both the AVM and VMM define methods for copying objects as part of the methodology. In Vera we had a built-in method called object_copy() as part of the language. This provided a deep copy of any object, without you having to write anything yourself. I found it useful, but the problem was that there was no control, so you could end up copying something *very* large.

Of course when writing custom copy functions we sometimes put bugs in them too.

Jason

]]> Comment on SystemVerilog Gotcha: (when copying) a struct is not a class by another name by Daniel Pörsch http://www.verilab.com/blog/2008/01/gotcha-when-copying-a-struct-is-not-a-class-by-a-different-name/#comment-154 Daniel Pörsch Tue, 22 Jan 2008 13:37:14 +0000 http://www.verilab.com/blog/2008/01/gotcha-when-copying-a-struct-is-not-a-class-by-a-different-name/#comment-154 Hi Verilab, just a short note, you define A,B as classes and X,Y as structs in the code, but refer to A,B as structs and X,Y as classes in the text. This is confusing, I guess the text is correct?! What would happen if I didn't have arrays of classes/structs but a single class and a single struct, I guess the same rule would apply, i.e. copy-by-reference for classes and copy-by-value for structs? There's no Copy-Constructor in SV, is there? Would be handy ... Rgds, Daniel Hi Verilab,

just a short note, you define A,B as classes and X,Y as structs in the code, but refer to A,B as structs and X,Y as classes in the text. This is confusing, I guess the text is correct?!

What would happen if I didn’t have arrays of classes/structs but a single class and a single struct, I guess the same rule would apply, i.e. copy-by-reference for classes and copy-by-value for structs?

There’s no Copy-Constructor in SV, is there? Would be handy …

Rgds,
Daniel

]]> Comment on You’ve Got [Mail|Bugs]? by Dusty http://www.verilab.com/blog/2007/12/youve-got-mailbugs/#comment-153 Dusty Sun, 16 Dec 2007 07:26:49 +0000 http://www.verilab.com/blog/2007/12/youve-got-mailbugs/#comment-153 E-mail can be used for to-do's with folders. Your inbox should be just that, an inbox. If it's informative, it can go into another folder. If it requires action, it goes in Todo. If it's fluff, scan it over and throw it in the trash. That's probably why using a to-do list or bug tracker for e-mail correspondence doesn't work as well. There's quite a bit of fluff that you get through e-mail. On the other side of things, tracking bugs through e-mail would limit all kinds of metrics and change tracking that management, bean counters, and excel graph afficiandos everywhere love. This is a big part of some process driven businesses. However, there's nothing to say you can't use a bug tracker centralized database with e-mail. Send your bug report to bugs@myproj and it could be tracked by the daemon reading it while sending an e-mail to the developer to look at it. It's all about how you format the text and if the daemon is smart enough to figure it all out so you aren't wasting more time trying to format your e-mail message. E-mail can be used for to-do’s with folders. Your inbox should be just that, an inbox. If it’s informative, it can go into another folder. If it requires action, it goes in Todo. If it’s fluff, scan it over and throw it in the trash. That’s probably why using a to-do list or bug tracker for e-mail correspondence doesn’t work as well. There’s quite a bit of fluff that you get through e-mail.

On the other side of things, tracking bugs through e-mail would limit all kinds of metrics and change tracking that management, bean counters, and excel graph afficiandos everywhere love. This is a big part of some process driven businesses. However, there’s nothing to say you can’t use a bug tracker centralized database with e-mail. Send your bug report to bugs@myproj and it could be tracked by the daemon reading it while sending an e-mail to the developer to look at it. It’s all about how you format the text and if the daemon is smart enough to figure it all out so you aren’t wasting more time trying to format your e-mail message.

]]> Comment on Reuseless Code by Sandeep Gor http://www.verilab.com/blog/2007/09/reuseless-code/#comment-152 Sandeep Gor Wed, 12 Dec 2007 08:57:52 +0000 http://www.verilab.com/blog/2007/09/reuseless-code/#comment-152 Nice explaination to my dilemma while developing verification environment. Lately, I started realizing what you explained as "reuseless code". In my last project, my verification environment implementation made me think "whether that perticular reusable code was really reusable? or was it required to implement it as resuable?" Thanks for this post. Best Regards, Sandeep Gor. Nice explaination to my dilemma while developing verification environment. Lately, I started realizing what you explained as “reuseless code”. In my last project, my verification environment implementation made me think “whether that perticular reusable code was really reusable? or was it required to implement it as resuable?”

Thanks for this post.

Best Regards,
Sandeep Gor.

]]> Comment on Checks or Functional Coverage (Part II)? by Sandeep Gor http://www.verilab.com/blog/2007/07/checks-or-fcov-2/#comment-151 Sandeep Gor Wed, 12 Dec 2007 06:08:42 +0000 http://www.verilab.com/blog/2007/07/checks-or-fcov-2/#comment-151 Hi David, I had a pleasur to work with you at Infineon's DVBH project. By the way, I fully agree with your comment that "checkers are more important than functional coverage". In fact, we need to have coverage points defined for checkers. Cadence have come up with a shareware to measure "check that" and "expect" statment and have them into coverage group. Specman 6.0 supports it. Once all checkers are developed, we should also have checker coverage grid along with functional coverage grid. Not only this, we should also make sure that each functional coverage point has relevent checker coverage point and each checker coverage point has relevent functional coverage point. This way, we make sure that all stimuli (functional coverage) has corrosponding output checker (checker coverage) in place. I have posted a small description about it on my blog http://digitalverification.blogspot.com/search/label/Coverage under heading (100% functional coverage is not enough). Best Regards, Sandeep Gor Hi David,
I had a pleasur to work with you at Infineon’s DVBH project.
By the way, I fully agree with your comment that “checkers are more important than functional coverage”. In fact, we need to have coverage points defined for checkers. Cadence have come up with a shareware to measure “check that” and “expect” statment and have them into coverage group. Specman 6.0 supports it.

Once all checkers are developed, we should also have checker coverage grid along with functional coverage grid. Not only this, we should also make sure that each functional coverage point has relevent checker coverage point and each checker coverage point has relevent functional coverage point. This way, we make sure that all stimuli (functional coverage) has corrosponding output checker (checker coverage) in place.

I have posted a small description about it on my blog http://digitalverification.blogspot.com/search/label/Coverage
under heading (100% functional coverage is not enough).

Best Regards,
Sandeep Gor

]]> Comment on DFT Digest: Secure Design-For-Test by John http://www.verilab.com/blog/2007/12/dft-digest-secure-design-for-test/#comment-150 John Thu, 06 Dec 2007 16:53:33 +0000 http://www.verilab.com/blog/2007/12/dft-digest-secure-design-for-test/#comment-150 Thanks for the hat tip JL - much appreciated. I'll surely return the favor. Having done a smidge of verification in my time, I find the Verilab (and CoolVerification) content interesting - and what's most interesting to me is that fact that it's grown so fast into something that is beyond my hardware-oriented aptitudes. 'e', SystemVerilog: OO is something I never got in school! I guess it's best left to the experts, like Verilab! Thanks for the hat tip JL - much appreciated. I’ll surely return the favor.

Having done a smidge of verification in my time, I find the Verilab (and CoolVerification) content interesting - and what’s most interesting to me is that fact that it’s grown so fast into something that is beyond my hardware-oriented aptitudes. ‘e’, SystemVerilog: OO is something I never got in school! I guess it’s best left to the experts, like Verilab!

]]>