Hierarchy view

In your documentation and tutorials you also speak about "design hierarchy". Normally the design is composed by multiple modules and I was expecting to see a real full design hierarchy where I have the top modules and then the other ones organized in a tree view.
I understand that once you open the file containing the top module you really have the full design hierachy, but once you navigate opening an inner module you loose that view, and the only way to go back is to click again in the tab containing the top module. It is extremaly frustrating if you need to browse inside a big project (with about 90 modules) not written by you...
It seems that the editor has only a file/module view and not a real design view... It should be useful to have the ability to lock the hierachy view to one file that should be designated as the top module so that it always reflect the entire design hierarchy.
I have seen a small undocumented triangle in the hierachy view that says "Select toplevel..." but the result I obtain is to loose the control of the view and have it empty once I change focus to another tab.

Another useful option should be to have also a filter in the hierarchycal view where we can hide objects like signals and processed and also the option to have a flat a-z ordered view where it easy to locate something inside of the entire design.

I have also found the "rename" feature extremely slow, considering a project with only 9 files! It took about 30 seconds before showing the preview window... I agree that it is anyway faster than manualy changing all the references, but I uses other refactor tool inside Visual Studio that are 100 time faster...

Also a last comment about port management in the entity. It seems to me that it is not possible to remove ports but only to add, and once you add it is added only at the end of the module... Not to say what it happens to the formatting...

Thank you anyway for having build an useful tool for writing VHDL...

Hierarchy view

Hi Enrico,

the behavior you describe for the hierarchy view is already implemented in our development version. It is part of in the upcoming release (scheduled for next week).

Hendrik.

Filter outline and hierarchy view

We have this logged as ticket:195
Any other supporters to prioritize this ticket?

Filter outline and hierarchy view

1 vote from me to give some priority to this feature

Filter hierarchy view

We discussed ticket:195 with the team and agreed to implement this in incremental steps.

As a first step we plan a "Show instantiations only" filter for the hierarchy view.

Hendrik.

The filter is a great idea

The filter is a great idea and I would like to have the same options for the outline view as well.

The first implementation, however, is a bit 'radical', because it also hides instantiations framed by the 'generate' statement.

Sorry for complaining again :-)
Kurt.

constructive complaining

Hi Kurt,

No worries. We do not mind a bit of constructive complaining. That is exactly the reason why we release new features "early".

I'll a note that we do want to show if-generate and for-generate if they contain an instantiation (or another generate statement that contains an instantiation). (logged as ticket:821)

Philippe

fixed

Hi Kurt,

We've iterated over the initial design and we have ticket:821 closed and ready to be released in next month's update.

Filtering instantiations is aware of generate statementsFiltering instantiations is aware of generate statements

Filter for outline view?

What is the current priority of ticket:195? It would be very nice to have filtering options for the outline view as well.

Btw. thanks for fixing the generate issue (ticket:821)

Regards,
Kurt.

feature specifications for outline filter

Hi Kurt,

You're welcome!

As for ticket:195, after we had implemented filters for the hierarchy, we had no immediate plans to work on the outline view filters.

However, this is surely negotiable ;-)

How would you use this feature? What would you like to filter out? Any other users want this?

-- Philippe

grouping of signal assignments

Hi Kurt,

Thanks for your e-mail with an example of your problem.
To summarize, you loose a lot of screen real-estate if you have multiple assignments to different bits of the same vector:

This would yield eight lines in the outline view. A 32 bit vector would be a terrible waste of screen space.

The first thing we can do is filter out all concurrent signal assignments. This would leave the more useful statements (mostly processes and instantiations) in the outline. I'm not a fan of filtering out everything but the instantiations. Can't put my finger on why I don't like, but it feels wrong.

Another path, but that would take longer to implement, is that we can try to group similar consecutive assignments into one entry in the outline view. In this example, we would have a single entry for "a" in the outline view. This would require some magic, but it is doable.

What do you think?

-- Philippe

grouping of signal assignments

Exactly, 'real estate' is the problem, although today's screens are bigger than before :-)

The outline view would be great to get an overview and to navigate quickly through a file. For my preferred editor I wrote a set of regular expressions, which show me constants, declarations, generates and instantiations all with different symbols in the outline view.

It would be great if one could select which elements to exclude. It would be pure luxury if shape and colour of the symbols are also configurable (but wisely chosen defaults would do for now :-)

I imagine that collecting assignments could make Sigasi slower, especially if they are not in sequence and distributed all over a file. But it would be an cool feature, also for problem checking (multiple assignments).

A better coding style would be the other solution, of course...

Kurt.

Slow rename?

Is it an option to send us your project, so I can profile was goes wrong?

Did you use the stand-alone version or the plugin? If you used the plugin, did you increase the default heap size settings of Eclipse?

Hendrik.

Port management

In future versions we indeed plan to implement more functionality to manage ports: delete, reorder, change datatype, ...

Slow rename!

In my project the “rename processor” is extremely slow. I have to wait for about 5 minutes until the name of a constant that is used 7 times in 5 files has changed. It doesn't matter whether I use the preview feature or not. This is very annoying, I've already spend the last 30 minutes waiting for the tool to do it's job.
I use the stand-alone version of HDT on a Windows 7 64 bit system.

speed

Hi Daniel,

I agree that five minutes is way too slow. I would like to reproduce your problem on our machines.

  • Can you tell me how many files are in your project (VHDL and non-VHDL)? Are there any files larger than 2000 lines?
  • Do you use a network drive (like SMB or a remotely shared directory)?
  • Does it take a similarly long time to perform a search operation on the same element (Search > Search references)?

We have tracked down a performance bug that will be fixed in the next release, but that bug might or might not be the cause of your problem.

Philippe

speed

Hi Philippe,
thank you for your fast reply.

* Can you tell me how many files are in your project (VHDL and non-VHDL)? Are there any files larger than 2000 lines?
I've removed a post-par simulation model vhdl-file (40000 lines). Now there are 68 files (RTL and structural VHDL) in 6 sub-folders and no file is larger than 2000 lines. Instead of 6 minutes a rename process now takes roughly 1min:30 sec.

* Do you use a network drive (like SMB or a remotely shared directory)?
No, all the files are located inside the project folder.

* Does it take a similarly long time to perform a search operation on the same element (Search > Search references)?
Yes

I've tried the rename feature on a Windows XP 32 bit system running on a VM.
Rename of the same element just took a few milliseconds. It seems to be an problem regarding Windows 7.

Kind Regards
Daniel

Hi Daniel, Thanks for your

Hi Daniel,

Thanks for your input. This will help me find the problem.

For now, I suggest that you mark the the simulation file as excluded. (right-click the file and select Exclude yourfile.vhd. We will work on other ways to speed up working with huge files in the future.

We can't blame Windows for this problem (even though I want to). Linux has a better performance when it comes to disk access. Perhaps the VM uses an extra caching mechanism, which increases the performance. It is possible that WinXP has the same performance problems if you run it on the "bare metal" (without VM).

I will investigate further.

kind regards

Philippe

speed

Hi Philippe,

It seems that you're right. I've tried the same rename process on a native Windows XP 32 bit running on the same machine. It showed the same slow performance as native Windows 7.

I hope this helps.

Kind Regard
Daniel

update

We have released some serious performance improvements in October of 2010: http://www.sigasi.com/content/performance-improvements

Post new comment

The content of this field is kept private and will not be shown publicly.
By submitting this form, you accept the Mollom privacy policy.