How about NessusClient 1.2-beta1 release?

[Jan-Oliver Wagner]

Hello,

CVS HEAD of NessusClient has a number of improvements
compared to version 1.0. I think it is already pretty stable
though I have still some improvements on my todo list.

Since noone except CVS-writers can try out the current
stuff I propose to make a "1.2-beta1" release. It can also be
a "1.1.0" release - whatever revisioning concept NessusClient
is going to follow.

How about such a beta release?
What needs to be done for such a release?

All the best

Jan

[Renaud Deraison]

Post by Jan-Oliver Wagner
Hello,
CVS HEAD of NessusClient has a number of improvements
compared to version 1.0. I think it is already pretty stable
though I have still some improvements on my todo list.
Since noone except CVS-writers can try out the current
stuff I propose to make a "1.2-beta1" release. It can also be
a "1.1.0" release - whatever revisioning concept NessusClient
is going to follow.
How about such a beta release?
What needs to be done for such a release?

Dedicated my morning to testing HEAD, and I don't think we're close
to a release yet. There are some regressions (ie: all the reports
indexes are recomputed every time nessusclient is loaded, which is
very slow if you use it to perform big scans), the GUI is very messy
(a lot of items need to be resized -- the scan list hbox is way to
wide compared to the scan configuration hbox which is 30 pixels wide
here), and the memory usage is abnormally high (130M here).

More effort needs to be done to make the GUI useable and to reduce
the memory usage, then we can do a beta. At the moment, doing a beta
would be a diservice to the users since there are known showstopper
with this program.



-- Renaud

[Jan-Oliver Wagner]

Post by Renaud Deraison

Post by Jan-Oliver Wagner
How about such a beta release?
What needs to be done for such a release?

Dedicated my morning to testing HEAD, and I don't think we're close
to a release yet. There are some regressions (ie: all the reports
indexes are recomputed every time nessusclient is loaded, which is
very slow if you use it to perform big scans), the GUI is very messy
(a lot of items need to be resized -- the scan list hbox is way to
wide compared to the scan configuration hbox which is 30 pixels wide
here), and the memory usage is abnormally high (130M here).
More effort needs to be done to make the GUI useable and to reduce
the memory usage, then we can do a beta. At the moment, doing a beta
would be a diservice to the users since there are known showstopper
with this program.

I am aware of these issues. Those are the ones that are on my list.
I would never release this status as "release candidate" - obviouly.
For me a beta is an intermediate state more or less close (ie feature
complete) to the next release. Known bugs are likely present.
Finding unknown early is a good thing.

Another problem I would like to solve with a beta
is to make available the HEAD code so that interested people
can work on the code. As mentioned already, there is currently
no other way to get it (only exception are the people with
CVS write access).

Perhaps a source-code only release will do the job?

Any other way to make the current HEAD code base available
would also help. E.g. configuring ViewCVS to allow tar-ball download.

Best

Jan

[Renaud Deraison]

Post by Jan-Oliver Wagner
I am aware of these issues. Those are the ones that are on my list.
I would never release this status as "release candidate" - obviouly.
For me a beta is an intermediate state more or less close (ie feature
complete) to the next release. Known bugs are likely present.
Finding unknown early is a good thing.

It is not feature complete : there are serious regressions in this
branch which may
cause scans contents to be lost.

Once this version is on par with 1.0 we may consider a beta release,
but at this point
it is way too premature.


-- Renaud

[Jan-Oliver Wagner]

Post by Renaud Deraison

Post by Jan-Oliver Wagner
I am aware of these issues. Those are the ones that are on my list.
I would never release this status as "release candidate" - obviouly.
For me a beta is an intermediate state more or less close (ie feature
complete) to the next release. Known bugs are likely present.
Finding unknown early is a good thing.

It is not feature complete : there are serious regressions in this
branch which may
cause scans contents to be lost.

I am using this version for a couple of month now and the only
situation where I lost scan contents was in situation where
memory consumption exceeded available memory. This is on my
list, but if you switch off the plugin caches in the preferences,
the behaviour should not differ from NessusClient 1.0.
Hm, maybe the caches should better be off by default?

If you lost information in a situation other than extreme memory consumption,
I'd be very interested in details.

Post by Renaud Deraison
Once this version is on par with 1.0 we may consider a beta release,
but at this point it is way too premature.

AFAICS, I am the only one to work towards maturity status.
Because I can only spend about three hours of my spare time per week,
it will take about a year to get all remaining issues done.

I am willing to motivate others to help to get things done faster.
But it is a major blocker that no one has the opportunity to get involved
with CVS HEAD. I'd appreciate any proposal how to improve this.

BTW: My current rough plan is:
- finish upgrade to GTK 2.4 ie. replace any deprecated methods.
- doing code review during this.
- Optimize widget/window sizes
- introduce some hashing regarding plugin data in order to decrease
memory consumption for cashing mode.

Best

Jan

[Renaud Deraison]

Post by Jan-Oliver Wagner
I am using this version for a couple of month now and the only
situation where I lost scan contents was in situation where
memory consumption exceeded available memory. This is on my
list, but if you switch off the plugin caches in the preferences,
the behaviour should not differ from NessusClient 1.0.
Hm, maybe the caches should better be off by default?

No, I can see a constantly growing memory usage.

Post by Jan-Oliver Wagner
If you lost information in a situation other than extreme memory consumption,
I'd be very interested in details.

But that's the point, isn't it ? What is in CVS can't handle a
moderately large scan, so I do not see the point into doing a
"release" that some distros might decide to include (as -devel) even
though it can cause harm to a lot of end-users who'll loose their
scans results.

There is absolutely no urgency in doing a release, and given the
state of the code there's no benefit other than wasting memory all
over the place.

I'll do a beta when I feel a beta is ready. There are things I want
to add in HEAD, as well as others I want to get rid of.

Now is absolutely not the time to release anything, the code is far
from being ready nor complete.


-- Renaud

[Jan-Oliver Wagner]

Post by Renaud Deraison

Post by Jan-Oliver Wagner
I am using this version for a couple of month now and the only
situation where I lost scan contents was in situation where
memory consumption exceeded available memory. This is on my
list, but if you switch off the plugin caches in the preferences,
the behaviour should not differ from NessusClient 1.0.
Hm, maybe the caches should better be off by default?

No, I can see a constantly growing memory usage.

Which actions trigger the increase?
Do you have all of the plugin caches switched off?

Post by Renaud Deraison

Post by Jan-Oliver Wagner
If you lost information in a situation other than extreme memory consumption,
I'd be very interested in details.

But that's the point, isn't it ? What is in CVS can't handle a
moderately large scan, so I do not see the point into doing a
"release" that some distros might decide to include (as -devel) even
though it can cause harm to a lot of end-users who'll loose their
scans results.

Honestly, I do not undestand: what point?
I simply was asking for details. If I don't get them, I can't work on the
problem.

Post by Renaud Deraison
There is absolutely no urgency in doing a release, and given the
state of the code there's no benefit other than wasting memory all
over the place.

My last email did not say it is urgent to have a release.
It basically said it would be good to have more people be able to
discuss technical issues as we currently discuss.

Post by Renaud Deraison
I'll do a beta when I feel a beta is ready.

OK.

Post by Renaud Deraison
There are things I want to add in HEAD,
as well as others I want to get rid of.

I'd be interested in what those things are, so
I might save time not working on things that are dropped
or deeply changed anyway.

Post by Renaud Deraison
Now is absolutely not the time to release anything, the code is far
from being ready nor complete.

OK. And I also do understand your fear that braindead distributors would
package even entirely buggy betas.

Do you actually mean not to give HEAD code to anyone in general
before it reaches beta quality?

Best

Jan