TimeStamp - can we remove references to this class?

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
5 messages Options
cbc
Reply | Threaded
Open this post in threaded view
|

TimeStamp - can we remove references to this class?

cbc
Hi. First, the background.

I've been living in (and using daily) an image dating back to roughly 4.10 times, but updated to some point post-5.1 , and lately it has become very clear that I have really done something nasty to the image.  When I save and quit, errors pop up, but it saves successfully and restarts without errors.
On the machine I saved it on.  I can't get the image to run on any other system that I have access too, which is scary.

I think this is related to the fact that I use ODBC, which has its own definition of TimeStamp.  This has collided significantly with the current Squeak definition of TimeStamp, and I think this collision has caused my these issues.

I'll be working on a new packaging of the ODBC client that doesn't reference TimeStamp at all, but rather deals with DateAndTime.  But that's just one package that used a 'TimeStamp' - I believe there are others out there.  Which brings up my question here:

Can we remove TimeStamp from the back package?

Its representation is definitely not what any older package would expect of TimeStamp.
TimeStamp appears to be a shim over DateAndTime for: TestRunner/SUnit, SqueakMap, Monticello, and ChangeRecord.  These could learn to use the DateAndTime, although that would require careful changes (very careful!).

Thanks,
cbc


Reply | Threaded
Open this post in threaded view
|

Re: TimeStamp - can we remove references to this class?

Chris Muller-3
TimeStamp does seem redundant, I see no reason we shouldn't get rid of
it.  Somebody just needs to do it (while taking the necessary care you
alluded to)..

On Wed, Mar 29, 2017 at 7:30 PM, Chris Cunningham
<[hidden email]> wrote:

> Hi. First, the background.
>
> I've been living in (and using daily) an image dating back to roughly 4.10
> times, but updated to some point post-5.1 , and lately it has become very
> clear that I have really done something nasty to the image.  When I save and
> quit, errors pop up, but it saves successfully and restarts without errors.
> On the machine I saved it on.  I can't get the image to run on any other
> system that I have access too, which is scary.
>
> I think this is related to the fact that I use ODBC, which has its own
> definition of TimeStamp.  This has collided significantly with the current
> Squeak definition of TimeStamp, and I think this collision has caused my
> these issues.
>
> I'll be working on a new packaging of the ODBC client that doesn't reference
> TimeStamp at all, but rather deals with DateAndTime.  But that's just one
> package that used a 'TimeStamp' - I believe there are others out there.
> Which brings up my question here:
>
> Can we remove TimeStamp from the back package?
>
> Its representation is definitely not what any older package would expect of
> TimeStamp.
> TimeStamp appears to be a shim over DateAndTime for: TestRunner/SUnit,
> SqueakMap, Monticello, and ChangeRecord.  These could learn to use the
> DateAndTime, although that would require careful changes (very careful!).
>
> Thanks,
> cbc
>
>
>

Reply | Threaded
Open this post in threaded view
|

Re: TimeStamp - can we remove references to this class?

Eliot Miranda-2
In reply to this post by cbc
Hi Chris,

   I support you and Chris in getting rid of the ODBC package's TimeStamp.  Is there a test suite?  Have you gone through the effects of relying on Squeak's TimeStamp?  Any good or bad news there?

On Wed, Mar 29, 2017 at 5:30 PM, Chris Cunningham <[hidden email]> wrote:
Hi. First, the background.

I've been living in (and using daily) an image dating back to roughly 4.10 times, but updated to some point post-5.1 , and lately it has become very clear that I have really done something nasty to the image.  When I save and quit, errors pop up, but it saves successfully and restarts without errors.
On the machine I saved it on.  I can't get the image to run on any other system that I have access too, which is scary.

I think this is related to the fact that I use ODBC, which has its own definition of TimeStamp.  This has collided significantly with the current Squeak definition of TimeStamp, and I think this collision has caused my these issues.

I'll be working on a new packaging of the ODBC client that doesn't reference TimeStamp at all, but rather deals with DateAndTime.  But that's just one package that used a 'TimeStamp' - I believe there are others out there.  Which brings up my question here:

Can we remove TimeStamp from the back package?

Its representation is definitely not what any older package would expect of TimeStamp.
TimeStamp appears to be a shim over DateAndTime for: TestRunner/SUnit, SqueakMap, Monticello, and ChangeRecord.  These could learn to use the DateAndTime, although that would require careful changes (very careful!).

Thanks,
cbc






--
_,,,^..^,,,_
best, Eliot


cbc
Reply | Threaded
Open this post in threaded view
|

Re: TimeStamp - can we remove references to this class?

cbc
Hi Chris and Elliot,
Thanks for the support. 

There are two requests here:
1. Remove TimeStamp from the image. After thinking more about this and especially looking on the web I noticed that timestamp iscrefetenced widely in several places in introductory texts and whatnot.  Maybe not the best idea to remove. But we could still have internal classes refer to DatAndTime instead.  I think that would be good. 
2.  Removing Timestamp from ODBC. There are no tests for this. There is a version where it was already removed in Squeaksource but this isn't on SqueakMap. And using the standard timestamp (or DateAndTime) for ODBC in general has some significant issues. I'll write about those once I get back to a computer. 

I think the next step is to have SqueakMap point to the package in Squeaksource, similar to what was done with FFI.

Thanks,
cbc 

On Mar 31, 2017 7:34 PM, "Eliot Miranda" <[hidden email]> wrote:
Hi Chris,

   I support you and Chris in getting rid of the ODBC package's TimeStamp.  Is there a test suite?  Have you gone through the effects of relying on Squeak's TimeStamp?  Any good or bad news there?

On Wed, Mar 29, 2017 at 5:30 PM, Chris Cunningham <[hidden email]> wrote:
Hi. First, the background.

I've been living in (and using daily) an image dating back to roughly 4.10 times, but updated to some point post-5.1 , and lately it has become very clear that I have really done something nasty to the image.  When I save and quit, errors pop up, but it saves successfully and restarts without errors.
On the machine I saved it on.  I can't get the image to run on any other system that I have access too, which is scary.

I think this is related to the fact that I use ODBC, which has its own definition of TimeStamp.  This has collided significantly with the current Squeak definition of TimeStamp, and I think this collision has caused my these issues.

I'll be working on a new packaging of the ODBC client that doesn't reference TimeStamp at all, but rather deals with DateAndTime.  But that's just one package that used a 'TimeStamp' - I believe there are others out there.  Which brings up my question here:

Can we remove TimeStamp from the back package?

Its representation is definitely not what any older package would expect of TimeStamp.
TimeStamp appears to be a shim over DateAndTime for: TestRunner/SUnit, SqueakMap, Monticello, and ChangeRecord.  These could learn to use the DateAndTime, although that would require careful changes (very careful!).

Thanks,
cbc






--
_,,,^..^,,,_
best, Eliot





Reply | Threaded
Open this post in threaded view
|

Re: TimeStamp - can we remove references to this class?

Dat Nguyen-2


Unsubscribe!


From: Squeak-dev <[hidden email]> on behalf of Chris Cunningham <[hidden email]>
Sent: Saturday, April 1, 2017 8:24 AM
To: The general-purpose Squeak developers list
Subject: Re: [squeak-dev] TimeStamp - can we remove references to this class?
 
Hi Chris and Elliot,
Thanks for the support. 

There are two requests here:
1. Remove TimeStamp from the image. After thinking more about this and especially looking on the web I noticed that timestamp iscrefetenced widely in several places in introductory texts and whatnot.  Maybe not the best idea to remove. But we could still have internal classes refer to DatAndTime instead.  I think that would be good. 
2.  Removing Timestamp from ODBC. There are no tests for this. There is a version where it was already removed in Squeaksource but this isn't on SqueakMap. And using the standard timestamp (or DateAndTime) for ODBC in general has some significant issues. I'll write about those once I get back to a computer. 

I think the next step is to have SqueakMap point to the package in Squeaksource, similar to what was done with FFI.

Thanks,
cbc 

On Mar 31, 2017 7:34 PM, "Eliot Miranda" <[hidden email]> wrote:
Hi Chris,

   I support you and Chris in getting rid of the ODBC package's TimeStamp.  Is there a test suite?  Have you gone through the effects of relying on Squeak's TimeStamp?  Any good or bad news there?

On Wed, Mar 29, 2017 at 5:30 PM, Chris Cunningham <[hidden email]> wrote:
Hi. First, the background.

I've been living in (and using daily) an image dating back to roughly 4.10 times, but updated to some point post-5.1 , and lately it has become very clear that I have really done something nasty to the image.  When I save and quit, errors pop up, but it saves successfully and restarts without errors.
On the machine I saved it on.  I can't get the image to run on any other system that I have access too, which is scary.

I think this is related to the fact that I use ODBC, which has its own definition of TimeStamp.  This has collided significantly with the current Squeak definition of TimeStamp, and I think this collision has caused my these issues.

I'll be working on a new packaging of the ODBC client that doesn't reference TimeStamp at all, but rather deals with DateAndTime.  But that's just one package that used a 'TimeStamp' - I believe there are others out there.  Which brings up my question here:

Can we remove TimeStamp from the back package?

Its representation is definitely not what any older package would expect of TimeStamp.
TimeStamp appears to be a shim over DateAndTime for: TestRunner/SUnit, SqueakMap, Monticello, and ChangeRecord.  These could learn to use the DateAndTime, although that would require careful changes (very careful!).

Thanks,
cbc






--
_,,,^..^,,,_
best, Eliot