Smalltalk › Pharo › Pharo Smalltalk Developers

Issue 3507 in pharo: SUnitUnloader >> unloadTestPackages is wrong and outdated

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

4 messages Options

Torsten Bergmann

Issue 3507 in pharo: SUnitUnloader >> unloadTestPackages is wrong and outdated

>I have the following remark: when I implemented this I took care to make all >remaining packages non-dirty. For example, you unload "Regex-Tests-Core" and >this will change the contents of the "Regex" package and hence make it >dirty. To avoid this, I moved tests into their own package – either into
>...
>Apparently somebody didn't take care when updating VB-Regex because the >tests are back in the main Regex package.

Hi Adrian,

I dont understand what you mean:

Anything we forgot was to update #unloadTestPackages (which is now
solved with issue 3507 by Mariano).

In Pharo1.2beta #12292 there is no "VB-Regex" or "Regex" package
anymore. So nothing to become dirty here.

There is a "Regex-Core", "Regex-Help" and "Regex-Tests-Core".
So test can easily be unloaded and existing packages are
still clean.

And yes ... if you would manually create a package "Regex"
then it would have become "dirty" (because class categories
are abused). But we know that and it can only be solved by
introducing real packages. Using a "RegexTest..." package
icould be a workaround and avoids this - but we use the
same naming scheme as Seaside.

Bye
T.

--
GMX DSL Doppel-Flat ab 19,99 Euro/mtl.! Jetzt mit
gratis Handy-Flat! http://portal.gmx.net/de/go/dsl

Adrian Lienhard

Re: Issue 3507 in pharo: SUnitUnloader >> unloadTestPackages is wrong and outdated

Hi Torsten,

Sorry, you are right. I apparently checked this in an 1.1.1 image (and I forgot that you have refactored Regex). Splitting the package up into -Core -Help and -Tests is a good solution indeed.

So the change of #unloadTestPackages proposed by Mariano makes perfect sense.

Cheers,
Adrian

On Jan 6, 2011, at 10:08 , Torsten Bergmann wrote:

>> I have the following remark: when I implemented this I took care to make all >remaining packages non-dirty. For example, you unload "Regex-Tests-Core" and >this will change the contents of the "Regex" package and hence make it >dirty. To avoid this, I moved tests into their own package – either into
>> ...
>> Apparently somebody didn't take care when updating VB-Regex because the >tests are back in the main Regex package.
>
> Hi Adrian,
>
> I dont understand what you mean:
>
> Anything we forgot was to update #unloadTestPackages (which is now
> solved with issue 3507 by Mariano).
>
> In Pharo1.2beta #12292 there is no "VB-Regex" or "Regex" package
> anymore. So nothing to become dirty here.
>
> There is a "Regex-Core", "Regex-Help" and "Regex-Tests-Core".
> So test can easily be unloaded and existing packages are
> still clean.
>
> And yes ... if you would manually create a package "Regex"
> then it would have become "dirty" (because class categories
> are abused). But we know that and it can only be solved by
> introducing real packages. Using a "RegexTest..." package
> icould be a workaround and avoids this - but we use the
> same naming scheme as Seaside.
>
> Bye
> T.
>
>
>
>
> --
> GMX DSL Doppel-Flat ab 19,99 Euro/mtl.! Jetzt mit
> gratis Handy-Flat! http://portal.gmx.net/de/go/dsl
>

Mariano Martinez Peck

Re: Issue 3507 in pharo: SUnitUnloader >> unloadTestPackages is wrong and outdated

On Thu, Jan 6, 2011 at 11:00 AM, Adrian Lienhard <[hidden email]> wrote:

Hi Torsten,

Sorry, you are right. I apparently checked this in an 1.1.1 image (and I forgot that you have refactored Regex). Splitting the package up into -Core -Help and -Tests is a good solution indeed.

So the change of #unloadTestPackages proposed by Mariano makes perfect sense.

Thanks for the remark anyway Adrian. Now, after evaluating the unloads, I open a Monticello Browser and I don't see any package as dirty. However, this is not really confident. Sometimes Monticello shows a package not direty, but just after doing a "changes" it gets dirty....is this enought or I should do something else to be sure there are no dirties?

thanks

mariano

Cheers,
Adrian

On Jan 6, 2011, at 10:08 , Torsten Bergmann wrote:

>> I have the following remark: when I implemented this I took care to make all >remaining packages non-dirty. For example, you unload "Regex-Tests-Core" and >this will change the contents of the "Regex" package and hence make it >dirty. To avoid this, I moved tests into their own package – either into
>> ...
>> Apparently somebody didn't take care when updating VB-Regex because the >tests are back in the main Regex package.
>
> Hi Adrian,
>
> I dont understand what you mean:
>
> Anything we forgot was to update #unloadTestPackages (which is now
> solved with issue 3507 by Mariano).
>
> In Pharo1.2beta #12292 there is no "VB-Regex" or "Regex" package
> anymore. So nothing to become dirty here.
>
> There is a "Regex-Core", "Regex-Help" and "Regex-Tests-Core".
> So test can easily be unloaded and existing packages are
> still clean.
>
> And yes ... if you would manually create a package "Regex"
> then it would have become "dirty" (because class categories
> are abused). But we know that and it can only be solved by
> introducing real packages. Using a "RegexTest..." package
> icould be a workaround and avoids this - but we use the
> same naming scheme as Seaside.
>
> Bye
> T.
>
>
>
>
> --
> GMX DSL Doppel-Flat ab 19,99 Euro/mtl.! Jetzt mit
> gratis Handy-Flat! http://portal.gmx.net/de/go/dsl
>

Adrian Lienhard

Re: Issue 3507 in pharo: SUnitUnloader >> unloadTestPackages is wrong and outdated

On Jan 6, 2011, at 17:54 , Mariano Martinez Peck wrote:

> Thanks for the remark anyway Adrian. Now, after evaluating the unloads, I
> open a Monticello Browser and I don't see any package as dirty. However,
> this is not really confident. Sometimes Monticello shows a package not
> direty, but just after doing a "changes" it gets dirty....is this enought or
> I should do something else to be sure there are no dirties?

Yes, there are indeed cases where the package is not correctly marked dirty.

I think that's safe to verify with "changes", though.

Cheers,
Adrian