* Clean out or vault old ncurses test versions
[not found] <169790885011.926668.9223018204115424627@server2.sourceware.org>
@ 2023-10-21 19:00 ` Brian Inglis
2023-10-21 20:15 ` Jon Turney
0 siblings, 1 reply; 6+ messages in thread
From: Brian Inglis @ 2023-10-21 19:00 UTC (permalink / raw)
To: cygwin-apps
On 2023-10-21 11:20, cygwin-no-reply@cygwin.com wrote:
> ERROR: x86_64/release/ncurses/libncurses++w10/libncurses++w10-6.4-4.20231016.tar.xz is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/libncurses++w10/libncurses++w10-6.4-4.20231016.hint is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/ncurses-demo/ncurses-demo-6.4-4.20231016.tar.xz is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/ncurses-demo/ncurses-demo-6.4-4.20231016.hint is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/terminfo-extra/terminfo-extra-6.4-4.20231016.tar.xz is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/terminfo-extra/terminfo-extra-6.4-4.20231016.hint is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/terminfo/terminfo-6.4-4.20231016.hint is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/terminfo/terminfo-6.4-4.20231016.tar.xz is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/ncurses-6.4-4.20231016.hint is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/ncurses-6.4-4.20231016-src.hint is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/ncurses-6.4-4.20231016-src.tar.xz is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/ncurses-6.4-4.20231016.tar.xz is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/ncurses-debuginfo/ncurses-debuginfo-6.4-4.20231016.tar.xz is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/ncurses-debuginfo/ncurses-debuginfo-6.4-4.20231016.hint is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/libncursesw10/libncursesw10-6.4-4.20231016.hint is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/libncursesw10/libncursesw10-6.4-4.20231016.tar.xz is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/libncurses-devel/libncurses-devel-6.4-4.20231016.tar.xz is both uploaded and automatically vaulted
> ERROR: x86_64/release/ncurses/libncurses-devel/libncurses-devel-6.4-4.20231016.hint is both uploaded and automatically vaulted
> ERROR: error while validating movelists for Brian Inglis
> SUMMARY: 19 ERROR(s)
Hi folks,
Thinking that ncurses i-i.net releases would pause at some point, I have been
incrementing the release number and appending the date throughout this year, but
it appears not, so I would now like to reset the primary release number to the
next after current stable 6.4-3.20230114 and use primary release 6.4-4.2023????
as my test prefix which I would like to make current stable some time soon!
Is there any way I can blow away my old test releases 6.4-5.2023... thru
6.4-13.2023... so I can reset the sequence, like listing a bunch of obsoletes
somehow:
6.4-5.20230514
6.4-6.20230520
6.4-7.20230603
6.4-8.20230617
6.4-9.20230625
6.4-10.20230701
6.4-11.20230708
6.4-12.20230715
6.4-13.20230729
or could someone kind person please do whatever is required if I can not do so?
--
Take care. Thanks, Brian Inglis Calgary, Alberta, Canada
La perfection est atteinte Perfection is achieved
non pas lorsqu'il n'y a plus rien à ajouter not when there is no more to add
mais lorsqu'il n'y a plus rien à retirer but when there is no more to cut
-- Antoine de Saint-Exupéry
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Clean out or vault old ncurses test versions
2023-10-21 19:00 ` Clean out or vault old ncurses test versions Brian Inglis
@ 2023-10-21 20:15 ` Jon Turney
2023-10-22 0:42 ` Brian Inglis
2023-10-22 15:42 ` Brian Inglis
0 siblings, 2 replies; 6+ messages in thread
From: Jon Turney @ 2023-10-21 20:15 UTC (permalink / raw)
To: Brian Inglis, cygwin-apps
On 21/10/2023 20:00, Brian Inglis via Cygwin-apps wrote:
> On 2023-10-21 11:20,
> cygwin-no-reply-rDBXBDvO6BXQT0dZR+AlfA@public.gmane.org wrote:
>> ERROR:
>> x86_64/release/ncurses/libncurses++w10/libncurses++w10-6.4-4.20231016.tar.xz is both uploaded and automatically vaulted
>> ERROR:
[..]
>> SUMMARY: 19 ERROR(s)
So, the reason why this confusing error is being emitted is because you
are trying to upload a test version 6.4-4.20231016, which is less than
the existing 6.4-12.20230715 and 6.4-13.20230729 test versions.
The default keep-count-test value is 2 (keep the two latest test
versions), so it would be expired immediately after upload.
This is an error, because that's not something sensible to do, and
usually indicates a mistake somewhere along the line.
> Thinking that ncurses i-i.net releases would pause at some point, I have
> been incrementing the release number and appending the date throughout
> this year, but it appears not, so I would now like to reset the primary
> release number to the next after current stable 6.4-3.20230114 and use
> primary release 6.4-4.2023???? as my test prefix which I would like to
> make current stable some time soon!
I'm quite clear on what this means, but this seems to be a problem of
upstream's making, if it really is releasing multiple versions called
"6.4" (with some date/patch level that isn't part of it's version label)
(e.g. look at https://repology.org/project/ncurses/information where
there's lots of variants on 6.4.x and no way to compare them because
individual distros make them up in different ways...)
> Is there any way I can blow away my old test releases 6.4-5.2023... thru
> 6.4-13.2023... so I can reset the sequence, like listing a bunch of
> obsoletes somehow:
The problem with the concept of "reset the sequence" is that
version-release identifiers have an ordering.
Anyone who already has these test releases installed, won't (ordinarily)
get downgraded to a lesser version.
> 6.4-5.20230514
> 6.4-6.20230520
> 6.4-7.20230603
> 6.4-8.20230617
> 6.4-9.20230625
> 6.4-10.20230701
> 6.4-11.20230708
> 6.4-12.20230715
> 6.4-13.20230729
>
> or could someone kind person please do whatever is required if I can not
> do so?
That said, you can use ssh vault command [1] to expunge versions that
are no longer required.
(and see the caveat there about how setup won't automatically downgrade
from removed versions)
If you really care about that, you could then upload appropriate
override.hint files (note that you need one per subpackage) with a
replace-version: line indicating the withdrawn version-release(s).
(It's unclear to me if this second step is really worth the effort,
given that only the presumably small number of people who install
ncurses test releases are affected.)
[1] https://cygwin.com/package-upload.html#deleting
Finally: you don't need to scrimp and save integers. I happen to have
an inexhaustible supply of monotonically increasing ones...
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Clean out or vault old ncurses test versions
2023-10-21 20:15 ` Jon Turney
@ 2023-10-22 0:42 ` Brian Inglis
2023-10-22 15:42 ` Brian Inglis
1 sibling, 0 replies; 6+ messages in thread
From: Brian Inglis @ 2023-10-22 0:42 UTC (permalink / raw)
To: cygwin-apps
On 2023-10-21 14:15, Jon Turney via Cygwin-apps wrote:
> On 21/10/2023 20:00, Brian Inglis via Cygwin-apps wrote:
>> On 2023-10-21 11:20, cygwin-no-reply-rDBXBDvO6BXQT0dZR+AlfA@public.gmane.org
>> wrote:
>>> ERROR:
>>> x86_64/release/ncurses/libncurses++w10/libncurses++w10-6.4-4.20231016.tar.xz
>>> is both uploaded and automatically vaulted
>>> ERROR:
> [..]
>>> SUMMARY: 19 ERROR(s)
>
> So, the reason why this confusing error is being emitted is because you are
> trying to upload a test version 6.4-4.20231016, which is less than the existing
> 6.4-12.20230715 and 6.4-13.20230729 test versions.
>
> The default keep-count-test value is 2 (keep the two latest test versions), so
> it would be expired immediately after upload.
>
> This is an error, because that's not something sensible to do, and usually
> indicates a mistake somewhere along the line.
>
>> Thinking that ncurses i-i.net releases would pause at some point, I have been
>> incrementing the release number and appending the date throughout this year,
>> but it appears not, so I would now like to reset the primary release number to
>> the next after current stable 6.4-3.20230114 and use primary release
>> 6.4-4.2023???? as my test prefix which I would like to make current stable
>> some time soon!
>
> I'm quite clear on what this means, but this seems to be a problem of upstream's
> making, if it really is releasing multiple versions called "6.4" (with some
> date/patch level that isn't part of it's version label)
>
> (e.g. look at https://repology.org/project/ncurses/information where there's
> lots of variants on 6.4.x and no way to compare them because individual distros
> make them up in different ways...)
>
>> Is there any way I can blow away my old test releases 6.4-5.2023... thru
>> 6.4-13.2023... so I can reset the sequence, like listing a bunch of obsoletes
>> somehow:
>
> The problem with the concept of "reset the sequence" is that version-release
> identifiers have an ordering.
>
> Anyone who already has these test releases installed, won't (ordinarily) get
> downgraded to a lesser version.
>
>> 6.4-5.20230514
>> 6.4-6.20230520
>> 6.4-7.20230603
>> 6.4-8.20230617
>> 6.4-9.20230625
>> 6.4-10.20230701
>> 6.4-11.20230708
>> 6.4-12.20230715
>> 6.4-13.20230729
>>
>> or could someone kind person please do whatever is required if I can not do so?
>
> That said, you can use ssh vault command [1] to expunge versions that are no
> longer required.
>
> (and see the caveat there about how setup won't automatically downgrade from
> removed versions)
>
> If you really care about that, you could then upload appropriate override.hint
> files (note that you need one per subpackage) with a replace-version: line
> indicating the withdrawn version-release(s).
>
> (It's unclear to me if this second step is really worth the effort, given that
> only the presumably small number of people who install ncurses test releases are
> affected.)
>
> [1] https://cygwin.com/package-upload.html#deleting
Thanks Jon,
I was unaware that useful command had been added.
> Finally: you don't need to scrimp and save integers. I happen to have an
> inexhaustible supply of monotonically increasing ones...
As does ncurses unfortunately; had I known they would be updating almost weekly,
I would have used the approach I am now going for, rather than the pattern
established earlier with less frequent updates.
--
Take care. Thanks, Brian Inglis Calgary, Alberta, Canada
La perfection est atteinte Perfection is achieved
non pas lorsqu'il n'y a plus rien à ajouter not when there is no more to add
mais lorsqu'il n'y a plus rien à retirer but when there is no more to cut
-- Antoine de Saint-Exupéry
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Clean out or vault old ncurses test versions
2023-10-21 20:15 ` Jon Turney
2023-10-22 0:42 ` Brian Inglis
@ 2023-10-22 15:42 ` Brian Inglis
2023-10-22 20:13 ` Jon Turney
1 sibling, 1 reply; 6+ messages in thread
From: Brian Inglis @ 2023-10-22 15:42 UTC (permalink / raw)
To: cygwin-apps
On 2023-10-21 14:15, Jon Turney via Cygwin-apps wrote:
> On 21/10/2023 20:00, Brian Inglis via Cygwin-apps wrote:
>> Thinking that ncurses i-i.net releases would pause at some point, I have been
>> incrementing the release number and appending the date throughout this year,
>> but it appears not, so I would now like to reset the primary release number to
>> the next after current stable 6.4-3.20230114 and use primary release
>> 6.4-4.2023???? as my test prefix which I would like to make current stable
>> some time soon!
> I'm quite clear on what this means, but this seems to be a problem of upstream's
> making, if it really is releasing multiple versions called "6.4" (with some
> date/patch level that isn't part of it's version label)
>
> (e.g. look at https://repology.org/project/ncurses/information where there's
> lots of variants on 6.4.x and no way to compare them because individual distros
> make them up in different ways...)
Thanks Jon,
The occasional/annual "official" GNU release is 6.4 but i-i.net seems to package
a new tarball with a date suffix every few days/weeks, after applying patches
developed or received.
The package naming pattern seems to be some variation of that with some
punctuation like ours but more fixed.
I should probably just skip the sequential "release" prefix to the date suffix,
as 6.4-2023???? is presumably greater than 6.4-?.2023????, and we have not yet
implemented EPOCH:V-R dating yet, correct?
>> Is there any way I can blow away my old test releases 6.4-5.2023... thru
>> 6.4-13.2023... so I can reset the sequence, like listing a bunch of obsoletes
>> somehow:
>> or could someone kind person please do whatever is required if I can not do so?
> That said, you can use ssh vault command [1] to expunge versions that are no
> longer required.
> (and see the caveat there about how setup won't automatically downgrade from
> removed versions)
> If you really care about that, you could then upload appropriate override.hint
> files (note that you need one per subpackage) with a replace-version: line
> indicating the withdrawn version-release(s).
> (It's unclear to me if this second step is really worth the effort, given that
> only the presumably small number of people who install ncurses test releases are
> affected.)
...and hopefully they ignore the sequential number.
Just to be clear about the process details and best practice:
I should create a dummy local dist tree for superseded versions,
with the tar file names prefixed with "-" and zero length,
and P-V-R.hint files replaced with override.hint files,
each containing "replace-versions: V-R..." for old versions,
then sftp upload that dist tree in a similar manner to cygport upload?
--
Take care. Thanks, Brian Inglis Calgary, Alberta, Canada
La perfection est atteinte Perfection is achieved
non pas lorsqu'il n'y a plus rien à ajouter not when there is no more to add
mais lorsqu'il n'y a plus rien à retirer but when there is no more to cut
-- Antoine de Saint-Exupéry
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Clean out or vault old ncurses test versions
2023-10-22 15:42 ` Brian Inglis
@ 2023-10-22 20:13 ` Jon Turney
2023-10-23 21:17 ` Brian Inglis
0 siblings, 1 reply; 6+ messages in thread
From: Jon Turney @ 2023-10-22 20:13 UTC (permalink / raw)
To: Brian Inglis, cygwin-apps
On 22/10/2023 16:42, Brian Inglis via Cygwin-apps wrote:
> I should probably just skip the sequential "release" prefix to the date
> suffix,
> as 6.4-2023???? is presumably greater than 6.4-?.2023????, and we have
> not yet implemented EPOCH:V-R dating yet, correct?
Yes, 2023???? (where ? stands for some digit) is bigger than any of the
numbers in the range 4...13.
Epochs are supported (theoretically), but I don't think you need to use
them here.
> ...and hopefully they ignore the sequential number.
Again, the question isn't about the meaning that a person ascribes to
the version number (because you can't reasonably expect people to
examine every package in detail), it's about the actions that setup will
automatically take.
> Just to be clear about the process details and best practice:
>
> I should create a dummy local dist tree for superseded versions,
> with the tar file names prefixed with "-" and zero length,
You don't need to do this. This removes files (which I think you have
already done)
> and P-V-R.hint files replaced with override.hint files,
> each containing "replace-versions: V-R..." for old versions,
> then sftp upload that dist tree in a similar manner to cygport upload?
Yes.
But if you are really going to upload a 6.4-2023MMDD sometime soon (or
anything greater than 6.4-13, in fact), you don't need to bother,
because that does supersede the removed test versions (and so any
installation of them will get upgraded).
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Clean out or vault old ncurses test versions
2023-10-22 20:13 ` Jon Turney
@ 2023-10-23 21:17 ` Brian Inglis
0 siblings, 0 replies; 6+ messages in thread
From: Brian Inglis @ 2023-10-23 21:17 UTC (permalink / raw)
To: cygwin-apps
On 2023-10-22 14:13, Jon Turney via Cygwin-apps wrote:
> On 22/10/2023 16:42, Brian Inglis via Cygwin-apps wrote:
>
>> I should probably just skip the sequential "release" prefix to the date suffix,
>> as 6.4-2023???? is presumably greater than 6.4-?.2023????, and we have not yet
>> implemented EPOCH:V-R dating yet, correct?
>
> Yes, 2023???? (where ? stands for some digit) is bigger than any of the numbers
> in the range 4...13.
>
> Epochs are supported (theoretically), but I don't think you need to use them here.
>
>> ...and hopefully they ignore the sequential number.
>
> Again, the question isn't about the meaning that a person ascribes to the
> version number (because you can't reasonably expect people to examine every
> package in detail), it's about the actions that setup will automatically take.
>
>> Just to be clear about the process details and best practice:
>>
>> I should create a dummy local dist tree for superseded versions,
>> with the tar file names prefixed with "-" and zero length,
>
> You don't need to do this. This removes files (which I think you have already
> done)
>
>> and P-V-R.hint files replaced with override.hint files,
>> each containing "replace-versions: V-R..." for old versions,
>> then sftp upload that dist tree in a similar manner to cygport upload?
>
> Yes.
>
> But if you are really going to upload a 6.4-2023MMDD sometime soon (or anything
> greater than 6.4-13, in fact), you don't need to bother, because that does
> supersede the removed test versions (and so any installation of them will get
> upgraded).
Thanks Jon,
I probably will do that in future.
For the inevitable future messes, wanted to build and test a script to make a
package obsolete with dummy files, which will replicate the current package dist
tree with each .hint replicated by replace-versions: ... override.hint and all
other (tar) files replaced with a zero length minus prefixed basename, uploaded
and readied by lftp.
I managed to run, test, and clean up, before my ISP connection went flakely.
--
Take care. Thanks, Brian Inglis Calgary, Alberta, Canada
La perfection est atteinte Perfection is achieved
non pas lorsqu'il n'y a plus rien à ajouter not when there is no more to add
mais lorsqu'il n'y a plus rien à retirer but when there is no more to cut
-- Antoine de Saint-Exupéry
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2023-10-23 21:17 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <169790885011.926668.9223018204115424627@server2.sourceware.org>
2023-10-21 19:00 ` Clean out or vault old ncurses test versions Brian Inglis
2023-10-21 20:15 ` Jon Turney
2023-10-22 0:42 ` Brian Inglis
2023-10-22 15:42 ` Brian Inglis
2023-10-22 20:13 ` Jon Turney
2023-10-23 21:17 ` Brian Inglis
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).