Please inspect: Migrated Trac tickets on our temporary GitHub server

[Matthias Koeppe]

We have just completed the first full trial migration of all ~35000 open and closed Trac tickets (as of late December) into a temporary GitHub Enterprise Server instance.

Please take a look at:

https://34.105.185.241/sagemath/sage-all-2023-01-12-003/issues

(accepting the self-signed server certificate)

The README of https://github.com/sagemath/trac-to-github explains some of the principles of the migration. It takes about 12 hours to get all data from Trac and to write a migration archive, and then about 20 hours to import it into our server. In the final migration (planned for Feb 1), our contact at GitHub.com will import a freshly generated migration archive into GitHub.com, and we will take down our temporary GitHub Enterprise Server instance.

Questions regarding the migration are very welcome. Please report any errors or wishlist items regarding the conversion here or at https://github.com/sagemath/trac-to-github/issues

[Dima Pasechnik]

don't get alarmed by clicking on a user and seeing "suspended" - it applies to this temporary server, not the real GitHub

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/fa7ccec5-4e81-4023-b9ed-4646f7e81754n%40googlegroups.com.

[Matthias Koeppe]

Next iteration: https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues

with much improved labels

[John Cremona]

On Mon, 16 Jan 2023 at 07:45, Matthias Koeppe <matthia...@gmail.com> wrote:

Next iteration: https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues

with much improved labels

The label component: elliptic curves is empty, which does not seem right.  On trac there are 64 open elliptic curve tickets.  One currently active example is #34782.  What is the best way to look for the new issue corresponding to a trac ticket number?

Currently the only authors on issues are Dima and Matthias.  Will it be possible to make the issue authors match the ticket creator on trac?

John

 

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/ee8d93b2-afe8-411a-b1ac-51586c0dc1d4n%40googlegroups.com.

[Dima Pasechnik]

On Mon, Jan 16, 2023 at 9:07 AM John Cremona <john.c...@gmail.com> wrote:
>
>
>
> On Mon, 16 Jan 2023 at 07:45, Matthias Koeppe <matthia...@gmail.com> wrote:
>>
>> Next iteration: https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues
>> with much improved labels
>
>
> The label component: elliptic curves is empty, which does not seem right. On trac there are 64 open elliptic curve tickets. One currently active example is #34782. What is the best way to look for the new issue corresponding to a trac ticket number?
>
> Currently the only authors on issues are Dima and Matthias.

No, why? See e.g.
https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues/21093
(author: you, John :-))
https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues/34898
(author: fchapoton)
https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues/34905
(author: tscrim)

> Will it be possible to make the issue authors match the ticket creator on trac?

the plan is that for everybody with a github account the trac tickets
acct names are mapped to
these account names. For the rest, they are mapped to some aliases.

The search on issues is currently not working, but it's probably a limitation of
our trial setup.

>
> John
>
>>
>>
>>
>> On Friday, January 13, 2023 at 11:26:50 AM UTC-8 Matthias Koeppe wrote:
>>>
>>> We have just completed the first full trial migration of all ~35000 open and closed Trac tickets (as of late December) into a temporary GitHub Enterprise Server instance.
>>>
>>> Please take a look at:
>>> https://34.105.185.241/sagemath/sage-all-2023-01-12-003/issues
>>> (accepting the self-signed server certificate)
>>>
>>> The README of https://github.com/sagemath/trac-to-github explains some of the principles of the migration. It takes about 12 hours to get all data from Trac and to write a migration archive, and then about 20 hours to import it into our server. In the final migration (planned for Feb 1), our contact at GitHub.com will import a freshly generated migration archive into GitHub.com, and we will take down our temporary GitHub Enterprise Server instance.
>>>
>>> Questions regarding the migration are very welcome. Please report any errors or wishlist items regarding the conversion here or at https://github.com/sagemath/trac-to-github/issues
>>>
>>>
>> --
>> You received this message because you are subscribed to the Google Groups "sage-devel" group.
>> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
>> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/ee8d93b2-afe8-411a-b1ac-51586c0dc1d4n%40googlegroups.com.
>
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAD0p0K4NYBt_p8xcdFY4xn-T4wBUUfMsGFNaDFKFawHv_EObkw%40mail.gmail.com.

[Dima Pasechnik]

On Mon, Jan 16, 2023 at 9:07 AM John Cremona <john.c...@gmail.com> wrote:
>
>
>

> On Mon, 16 Jan 2023 at 07:45, Matthias Koeppe <matthia...@gmail.com> wrote:
>>
>> Next iteration: https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues
>> with much improved labels
>
>
> The label component: elliptic curves is empty, which does not seem right. On trac there are 64 open elliptic curve tickets. One currently active example is #34782. What is the best way to look for the new issue corresponding to a trac ticket number?

Please re-check. Clicking on "elliptic curves" label gives me

https://34.105.185.241/sagemath/sage-all-2023-01-14-004/labels/component%3A%20elliptic%20curves

with 64 open tickets, and 410 closed.

[Dima Pasechnik]

On Mon, Jan 16, 2023 at 11:55 AM Dima Pasechnik <dim...@gmail.com> wrote:
>
> On Mon, Jan 16, 2023 at 9:07 AM John Cremona <john.c...@gmail.com> wrote:
> >
> >
> >
> > On Mon, 16 Jan 2023 at 07:45, Matthias Koeppe <matthia...@gmail.com> wrote:
> >>
> >> Next iteration: https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues
> >> with much improved labels
> >
> >
> > The label component: elliptic curves is empty, which does not seem right. On trac there are 64 open elliptic curve tickets. One currently active example is #34782. What is the best way to look for the new issue corresponding to a trac ticket number?

trac ticket XYZ corresponds to
https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues/XYZ

[John Cremona]

Thanks, Dima.  I do now see 64 issues when I first view all issues and then filter on that, so I don't know what was happening earlier.

As for authors: when I am looking at all issues and then click on the "author" tab, there are only 2 authors in the drop-down list.

John

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAAWYfq3%2BFe8UBbg%3DcDg_qMwDJ5OTWBX%3DKQmYOhLiJNjU95AE6w%40mail.gmail.com.

[Dima Pasechnik]

On Mon, Jan 16, 2023 at 12:22 PM John Cremona <john.c...@gmail.com> wrote:
>
> Thanks, Dima. I do now see 64 issues when I first view all issues and then filter on that, so I don't know what was happening earlier.
>
> As for authors: when I am looking at all issues and then click on the "author" tab, there are only 2 authors in the drop-down list.

that's probably because on the trial server only mkoeppe and dimpase
are active accounts.
As you see from individual issues, it's OK there.

Also, try in the issue's Filters tab to enter

is:open is:issue author:JohnCremona

This gives me 9 open and 248 closed issues.

>
> John
>
> On Mon, 16 Jan 2023 at 11:59, Dima Pasechnik <dim...@gmail.com> wrote:
>>
>> On Mon, Jan 16, 2023 at 9:07 AM John Cremona <john.c...@gmail.com> wrote:
>> >
>> >
>> >
>> > On Mon, 16 Jan 2023 at 07:45, Matthias Koeppe <matthia...@gmail.com> wrote:
>> >>
>> >> Next iteration: https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues
>> >> with much improved labels
>> >
>> >
>> > The label component: elliptic curves is empty, which does not seem right. On trac there are 64 open elliptic curve tickets. One currently active example is #34782. What is the best way to look for the new issue corresponding to a trac ticket number?
>>
>> Please re-check. Clicking on "elliptic curves" label gives me
>>
>> https://34.105.185.241/sagemath/sage-all-2023-01-14-004/labels/component%3A%20elliptic%20curves
>>
>> with 64 open tickets, and 410 closed.
>>
>> --
>> You received this message because you are subscribed to the Google Groups "sage-devel" group.
>> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
>> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAAWYfq3%2BFe8UBbg%3DcDg_qMwDJ5OTWBX%3DKQmYOhLiJNjU95AE6w%40mail.gmail.com.
>
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAD0p0K4%3Dx2KCr0dY8MYiHr7tgtfoUuC2hc1W6kKDPzwQwNs2iw%40mail.gmail.com.

[John Cremona]

On Mon, 16 Jan 2023 at 12:35, Dima Pasechnik <dim...@gmail.com> wrote:

On Mon, Jan 16, 2023 at 12:22 PM John Cremona <john.c...@gmail.com> wrote:
>
> Thanks, Dima.  I do now see 64 issues when I first view all issues and then filter on that, so I don't know what was happening earlier.
>
> As for authors: when I am looking at all issues and then click on the "author" tab, there are only 2 authors in the drop-down list.

that's probably because on the trial server only mkoeppe and dimpase
are active accounts.
As you see from individual issues, it's OK there.

Also, try in the issue's Filters tab to enter

  is:open is:issue author:JohnCremona

This gives me 9 open and 248 closed issues.

Yes!  I was wrongly looking for my old trac id (cremona).   

This all looks excellent to me, I am looking forward to the change.

 

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAAWYfq1T36uShK4FR3tQ7zwH1XMHwz1xqNie-89kRrAd01y-yg%40mail.gmail.com.

[Dima Pasechnik]

I don't know. Perhaps Matthias changed something in the setup of the repos.

On Mon, Jan 16, 2023 at 3:26 PM Aram Dermenjian <> wrote:
>
> Hey Dima,
>
> I'm confused by what you mean by "The search on issues is currently not working, but it's probably a limitation of
> our trial setup."
>
> When I tried the search on the earlier version it was working fine, but it's not working on this new version.
> https://34.105.185.241/sagemath/sage-all-2023-01-12-003/issues?q=34782+
> vs
> https://34.105.185.241/sagemath/sage-all-2023-01-14-004/issues?q=34782+
>
> -Aram

>> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAAWYfq1YHtUty5UjJo-LPzvLRQ04d1NF071oupre1h_gCsY9MA%40mail.gmail.com.

[Dima Pasechnik]

On Mon, Jan 16, 2023 at 3:34 PM Dima Pasechnik <dim...@gmail.com> wrote:
>
> I don't know. Perhaps Matthias changed something in the setup of the repos.

I doubt it's some sort of metadata poisoning that happened in the
latest iteration so that the search broke - it's probably general
flakiness of setup.
We can, potentially, ask GitHub support for help.
However, the Enterprise Server version we're using is not the very
latest - but I don't want more hours of banging my head on the
keyboard to upgrade it to the next minor version (as they'd probably
1st thing ask for such an update...)

[John Cremona]

I have spent some time inspecting the (provisionally) migrated tickets, and it all looks very good.  One small thing struck me: the cross-referencing to other tickets works fine (ticket/issue numbers preceded by "#" are links to the other ticket/issue), but one cannot tell without following the link whether the other ticket is closed.  This is important for other tickets which are dependencies.  On trac, these were displayed with a strike-through.  Is it possible to do the same, or automatically  append " (closed)" after a link to another closed ticket?  Or will this be unnecessary once github does its usual thing when tickets are cross-referenced?

Also, I think it is fine to have many different labels for issues, as we have many components.  But why was it thought necessary to have all the component labels start with "component:"?  That seems redundant. It does mean that all the components come almost first in the list of labels ("bug" is at the top which is fine), with the dozen or so non-component labels other than bug all at the bottom of the list.  If we could prefix all the non-component labels instead with something very small which was allowed by GitHub and which made them all come first, then we could remove the "component:" prefix from the majority.

Is it possible for us to try out creating a new (dummy) issue on the mock GitHub site, to try out the interface? Obviously any issues created there now will be lost at the next iteration.

John

To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAAWYfq2Sp6_sh%2B488jj%3D-_t3Mrgp9%3DODTmyw5i2PGd2X%2B4QV-g%40mail.gmail.com.

[Matthias Koeppe]

On Tuesday, January 17, 2023 at 2:10:24 AM UTC-8 john.c...@gmail.com wrote:

One small thing struck me: the cross-referencing to other tickets works fine (ticket/issue numbers preceded by "#" are links to the other ticket/issue), but one cannot tell without following the link whether the other ticket is closed.  This is important for other tickets which are dependencies.  On trac, these were displayed with a strike-through.  Is it possible to do the same, or automatically  append " (closed)" after a link to another closed ticket?

We won't have control over this for future issues opened on GitHub; so probably we shouldn't try to add this information for the migrated tickets.

One does not actually have to follow the link though; it's enough to hover over it, and a popup will show the status of the referenced issue.

 

[Matthias Koeppe]

On Tuesday, January 17, 2023 at 2:10:24 AM UTC-8 john.c...@gmail.com wrote:

I think it is fine to have many different labels for issues, as we have many components.  But why was it thought necessary to have all the component labels start with "component:"?  That seems redundant. It does mean that all the components come almost first in the list of labels ("bug" is at the top which is fine), with the dozen or so non-component labels other than bug all at the bottom of the list.  If we could prefix all the non-component labels instead with something very small which was allowed by GitHub and which made them all come first, then we could remove the "component:" prefix from the majority.

Controlling the sort order in the drop down list of the Issues UI was the reason for using label prefixes. That's why we have "component: " as well as "p1 –".... "p5 – " for the priority.

Tobias' initial proposal was to use the single letter "c: " as the prefix for components. I thought it would be too cryptic, so I went with the whole word "component: ". But I'd be happy to change it if "c: " (or some other idea) seems preferable to others.

 

[Matthias Koeppe]

On Tuesday, January 17, 2023 at 2:10:24 AM UTC-8 john.c...@gmail.com wrote:

Is it possible for us to try out creating a new (dummy) issue on the mock GitHub site, to try out the interface? Obviously any issues created there now will be lost at the next iteration.

Dima would be able to create an account that allows logins.

 

[Matthias Koeppe]

On Tuesday, January 17, 2023 at 2:10:24 AM UTC-8 john.c...@gmail.com wrote:

one cannot tell without following the link whether the other ticket is closed.  This is important for other tickets which are dependencies.

For the case of dependencies, there is a discussion in https://github.com/sagemath/trac-to-github/issues/78 with some pointers to bots (GH Actions workflows) that would keep track of which declared dependencies are still open. I have chosen the format in which the dependencies appear in the migrated ticket in a way that is compatible with the major candidates for such workflows.

[John Cremona]

Thanks Matthias for the detailed replies.  I, and I'm sure all other sage developers, are extremely grateful for the work you and the others have done for this migration.

John

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/5ea22368-93d5-4121-99fe-67628a780888n%40googlegroups.com.

[Matthias Koeppe]

Regarding label prefixes, I just tried a minimalist version, using just ": " as the prefix in https://34.105.185.241/sagemath/sage-20230116172535/issues, but unfortunately it does not work: The nonalphabetic prefix is ignored for sorting.

 

[Matthias Koeppe]

Also alphabetic non-ASCII unicode characters are unfortunately ignored for sorting, so using something like ⓒ and ⓟ as a prefix symbol does not work.

[Kwankyu Lee]

On Wednesday, January 18, 2023 at 4:08:15 AM UTC+9 Matthias Koeppe wrote:

Tobias' initial proposal was to use the single letter "c: " as the prefix for components. I thought it would be too cryptic, so I went with the whole word "component: ". But I'd be happy to change it if "c: " (or some other idea) seems preferable to others.

That was me, perhaps. Now I advocate for no prefix at all. It seems a futile attempt to sort labels in the list. It's better to make the labels (and their colors) look good and read well.

[Kwankyu Lee]

In the issue description, I see "Issue created by migration from https://trac.sagemath.org/ticket/...".  Would we remove this before migration? Perhaps we should because the link will be defunct eventually.

[Dima Pasechnik]

On Wed, Jan 18, 2023 at 8:16 AM Kwankyu Lee <ekwa...@gmail.com> wrote:
>
> In the issue description, I see "Issue created by migration from https://trac.sagemath.org/ticket/...". Would we remove this before migration? Perhaps we should because the link will be defunct eventually.

The plan is to have the trac website in a read-only form, perhaps on
archive,org,
or somewhere else. So I think it's OK to keep these - perhaps not clickable.
Anyhow, this can be edited later using a GitHub API.

>
>
>
>
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/9b560c7a-3f2f-4447-b2f5-ff580bb8ff09n%40googlegroups.com.

[seb....@gmail.com]

Since we can't be sure that all tickets have been converted correctly, I prefer to keep the link clickable for as long as possible! Furthermore, I think we should not erase the traces of the migration.

[Kwankyu Lee]

On Thursday, January 19, 2023 at 2:07:36 AM UTC+9 seb....@gmail.com wrote:

Since we can't be sure that all tickets have been converted correctly, I prefer to keep the link clickable for as long as possible! Furthermore, I think we should not erase the traces of the migration.

As long as trac is running, we can find the ticket by the issue number, which is the same with the ticket number.

After migration, we will care less with the welfare of trac. I think the links will always be at the risk of getting defunct.

 

[Matthias Koeppe]

In https://github.com/sagemath/trac-to-github/issues/80, I propose to set up http permanent redirects from trac.sagemath.org to corresponding GitHub URLs. 

When users follow a link to trac.sagemath.org from some external site and get redirected to the GH Issue, the message "Issue created by migration from..." reassures the user that they have arrived at the right place.

[Matthias Koeppe]

In the latest iteration, https://34.105.185.241/sagemath/sage-all-2023-01-14-009/issues, I have switched to the shorter prefix "c: " for components.

Regarding the component labels, we have an ongoing discussion in https://github.com/sagemath/trac-to-github/issues/99

[Kwankyu Lee]

There are duplicates of the same comments. See many comments below from

https://34.105.185.241/sagemath/sage-all-2023-01-14-012/issues/34909#issuecomment-4477657

[Matthias Koeppe]

I may have forgotten to run clear out the previous migration archive on this test import.

A new test import will be ready in a bit, I'll post the link when it's ready.

[Matthias Koeppe]

The latest iteration: https://34.105.185.241/sagemath/sage-all-2023-01-14-013/issues

(with some refinements on labels).

[Kwankyu Lee]

I am not sure if the "/ 1", "/ 2", "/ 3", "/ 4", "/ 5" postfixes are really helpful. Note that

(1) As Matthias noticed, the priority labels are sorted in the correct order without those prefixes. 

(2) The label text and the label color already indicate the priority.

(3) The numeral postfix is somewhat confusing since "5" may be interpreted as the highest priority and "1" lowest.

Matthias pointed out that we can choose the priority label by one key stroke (for example, you can type "1" to get "p: blocker / 1"). But why we insist to have the convenience that is absent for all other labels? 

[seb....@gmail.com]

But why we insist to have the convenience that is absent for all other labels?

How about doing it the other way round? Use it for the other labels too. Why not do it completely without prefixes and use the /abc postfix to have convienient access to all labels. For example algebra /ca in stead of c: algebra, algebraic geometry /cag instead of c: algebraic geometry … Typing ca will give you all component labels starting with a and cat pulls algebraic topology to the first row.

​

[seb....@gmail.com]

Sorry, Google killed the formatting of my Markdown plugin and I can't figure out how to edit the post any more. I hope you can read my preceding post!

[Matthias Koeppe]

On Wednesday, January 25, 2023 at 12:55:59 PM UTC-8 seb....@gmail.com wrote:

Why not do it completely without prefixes and use the /abc postfix to have convienient access to all labels. For example algebra /ca in stead of c: algebra, algebraic geometry /cag instead of c: algebraic geometry … 

We are using the prefixes so that in the alphabetically sorted label list in the user interface, there is a consecutive chunk of labels that correspond to components. It's very untidy when the list of component labels is interrupted by other labels such as "enhancement", "needs work" etc.

 

[seb....@gmail.com]

I knew that argument. But isn't it much easier to select an entry from a list of almost 100 entries by typing two or three characters instead of scrolling through the list? If we were to change enhancement to enhancement /te (t for type to Trac) and needs work to needs work /snw (s for status), these could be easily selected as well. Here I assume that it is not a problem to change a predefined label.

[Matthias Koeppe]

Typing "enh" selects "enhancement" uniquely

Typing "s w" selects "needs work" uniquely

Typing "s r" selects "needs review" uniquely

Typing "geo" gives you 2 options for "geometry" or "algebraic geometry" to select

etc.

So I don't think we need to add three-letter shortcuts, which would come with a cost of obscurity 

[Matthias Koeppe]

(Hopefully) final iteration: https://34.105.185.241/sagemath/sage-all-2023-01-14-014/issues

(with some fixes in the markdown conversion, fixes for comment numbers, and sorted description attributes)

[Matthias Koeppe]

(Final + 1)st iteration: https://34.105.185.241/sagemath/sage-all-2023-01-14-015/issues

(see https://github.com/sagemath/trac-to-github/commits/master)

[Matthias Koeppe]

(Final + 2)nd iteration: https://34.105.185.241/sagemath/sage-all-2023-01-14-016

[Matthias Koeppe]

Very final iteration: https://34.105.185.241/sagemath/sage-prod-2023-01-30-077/issues

(with the final data from the read-only Trac)