Drafting Tender "Cleanup & further improve ODF conformance"

Hello,

one of the approved [1] tenders is the

  Tender "Cleanup & further improve ODF conformance"

The board would like to work together in public with all of you on this tender before it gets officially published. The current draft is therefore shared at

  https://nextcloud.documentfoundation.org/s/ggqpciBK54rztJi

The board is happy to get your feedback and proposals. We'd like to discuss this ideally in the board call after next, i.e. on Friday, November 19, at 1300 Berlin time. Please send your feedback to the public board-discuss@documentfoundation.org mailing list.

Thanks a lot to all community members involved for working on the tender draft!

Looking forward to your feedback,
Florian

[1] https://listarchives.documentfoundation.org/www/board-discuss/2021/msg00091.html

Hi Florian,

(1)
The link http://autotests.opendocumentformat.org from item "Required 1." does not work.
Do you have another reference for ODFAutoTests?

(2)
The search result from item "Required 2." contains Meta-issues. Expanding them results in 80 issues.

Using Whiteboard as search criteria has no advantage compared to the Meta-issues. And I think both, Whitheboard search or Meta-issues, are not suitable for a tender, but a tender needs to list the issues explicitly.

The list from Whiteboard search and Meta-issues needs to be examined and prioritized manually.

(3)
Is it possible to get
https://bugs.documentfoundation.org/show_bug.cgi?id=48392
ODF: Implementation for svg:linearGradient and svg:radialGradient is missing
as explicit issue for "Required"?
We had this already as suggestion "Multi-color gradient" in
https://wiki.documentfoundation.org/Development/Budget2021
and now again in
https://wiki.documentfoundation.org/Development/Budget2022

Kind regards,
Regina

the git repo is here:

https://gitlab.com/odfplugfest/odfautotests

but it does look quite inactive, with the last commit in 2015.

possibly Jos van den Oever knows more...

the scope of this is quite large and unclear... *required* items are:

1. ODFAutoTests: addressing issues will be difficult because as Regina points out the web service appears to be offline.
    IIRC it's possible to run the tests offline, but currently i guess nobody knows how much work it is to set that up and what problems would actually be found, so i guess this item mostly amounts to "get ODFAutoTests to run at all".

2. odf_validation:

* 37128 this is, errm, "interesting" problem and might take weeks to fix
* 96066 likely needs specification work
* 94768 cannot be solved with ODF 1.3, it needs specification work
* 106934 needs specification work, possibly it was already added for ODF 1.4
* 131127 might be fixable?
* 131148 needs specification work
* 131159 this was added for ODF 1.4
* 108198 export meta-bug depending on 26 unfixed bugs, wow...
* 94587 *import* meta-bug depending on 37 unfixed bugs
   - how does this have "odf_validation" keyword in the first place,
     i thought that applied only to the export filter?
     i would propose to remove "odf_validation" keyword and keep "odf".

... so i'm not sure what would make sense here, certainly *requiring* fixes of > 60 different bugs that are all over the map doesn't make sense to me, unless the board wants to spend the entire yearly budget...

maybe everything should be "optional" and then applicants can list which bugs they think are actually possible to fix given the current ODF 1.3 specification?

Hello,

thanks a lot for the feedback, and sorry also here for the belated reply!

(1)
The link http://autotests.opendocumentformat.org from item "Required 1." does not work.
Do you have another reference for ODFAutoTests?

unfortunately not - I can confirm the website is not loading, so I'll replace the reference with the Git repo pointed out by Michael Stahl in the meantime.

(2)
The search result from item "Required 2." contains Meta-issues. Expanding them results in 80 issues.

Using Whiteboard as search criteria has no advantage compared to the Meta-issues. And I think both, Whitheboard search or Meta-issues, are not suitable for a tender, but a tender needs to list the issues explicitly.

The list from Whiteboard search and Meta-issues needs to be examined and prioritized manually.

This is taken from the specification at https://wiki.documentfoundation.org/Development/Budget2021#Cleanup_.26_further_improve_ODF_conformance

I fear answering that question is beyond my skills. :wink: Does it make sense to bounce this question back to the ESC for further specification?

(3)
Is it possible to get
https://bugs.documentfoundation.org/show_bug.cgi?id=48392
ODF: Implementation for svg:linearGradient and svg:radialGradient is missing
as explicit issue for "Required"?
We had this already as suggestion "Multi-color gradient" in
https://wiki.documentfoundation.org/Development/Budget2021
and now again in
https://wiki.documentfoundation.org/Development/Budget2022

I've added it. Not sure, however, how much that would change the work/cost estimate of the tender.

the scope of this is quite large and unclear... *required* items are:

1. ODFAutoTests: addressing issues will be difficult because as Regina points out the web service appears to be offline.
   IIRC it's possible to run the tests offline, but currently i guess nobody knows how much work it is to set that up and what problems would actually be found, so i guess this item mostly amounts to "get ODFAutoTests to run at all".

I've tried to rephrase #1 a bit, let me know if this is better.

2. odf_validation:

* 37128 this is, errm, "interesting" problem and might take weeks to fix
* 96066 likely needs specification work
* 94768 cannot be solved with ODF 1.3, it needs specification work
* 106934 needs specification work, possibly it was already added for ODF 1.4
* 131127 might be fixable?
* 131148 needs specification work
* 131159 this was added for ODF 1.4
* 108198 export meta-bug depending on 26 unfixed bugs, wow...
* 94587 *import* meta-bug depending on 37 unfixed bugs
  - how does this have "odf_validation" keyword in the first place,
    i thought that applied only to the export filter?
    i would propose to remove "odf_validation" keyword and keep "odf".

... so i'm not sure what would make sense here, certainly *requiring* fixes of > 60 different bugs that are all over the map doesn't make sense to me, unless the board wants to spend the entire yearly budget...

maybe everything should be "optional" and then applicants can list which bugs they think are actually possible to fix given the current ODF 1.3 specification?

Given the amount of changes from the original tender, I wonder if it makes sense to bounce this back to the ESC and discuss it in one of the next calls? We have the autoupdater tender discussion tomorrow already, so maybe it's a bit on too short notice and would fill the agenda too much, but it could be a topic in one of the next calls? Or is it better to suggest the above list via e-mail and use it if people are fine with it?

Thanks,
Florian

Hello everyone,

the below mail is a bit older - Christmas break and some other tenders came in between, so I get to this only now.

(1)
The link http://autotests.opendocumentformat.org from item "Required 1." does not work.
Do you have another reference for ODFAutoTests?

unfortunately not - I can confirm the website is not loading, so I'll replace the reference with the Git repo pointed out by Michael Stahl in the meantime.

This is done in https://nextcloud.documentfoundation.org/s/ggqpciBK54rztJi

(2)
The search result from item "Required 2." contains Meta-issues. Expanding them results in 80 issues.

Using Whiteboard as search criteria has no advantage compared to the Meta-issues. And I think both, Whitheboard search or Meta-issues, are not suitable for a tender, but a tender needs to list the issues explicitly.

The list from Whiteboard search and Meta-issues needs to be examined and prioritized manually.

This is taken from the specification at https://wiki.documentfoundation.org/Development/Budget2021#Cleanup_.26_further_improve_ODF_conformance

I fear answering that question is beyond my skills. :wink: Does it make sense to bounce this question back to the ESC for further specification?

Regina (thanks a lot!) sent a list of bugs back in December on the dev mailing list: https://lists.freedesktop.org/archives/libreoffice/2021-December/088210.html

Was there any further discussion or feedback on this? If the list mentioned there is fine, I replace item 2 from the tender with it. If we're unsure whether that meets the budget or not, as the person days are listed in the tender, we can add a note along the lines of "Please propose a subset and prioritization of these bugs, that do not exceed the person days factored in for this tender, see below."

(3)
Is it possible to get
https://bugs.documentfoundation.org/show_bug.cgi?id=48392
ODF: Implementation for svg:linearGradient and svg:radialGradient is missing
as explicit issue for "Required"?
We had this already as suggestion "Multi-color gradient" in
https://wiki.documentfoundation.org/Development/Budget2021
and now again in
https://wiki.documentfoundation.org/Development/Budget2022

I've added it. Not sure, however, how much that would change the work/cost estimate of the tender.

This is in the draft. We can add a similar note as mentioned above if we're unsure about the work required.

the scope of this is quite large and unclear... *required* items are:

1. ODFAutoTests: addressing issues will be difficult because as Regina points out the web service appears to be offline.
   IIRC it's possible to run the tests offline, but currently i guess nobody knows how much work it is to set that up and what problems would actually be found, so i guess this item mostly amounts to "get ODFAutoTests to run at all".

I've tried to rephrase #1 a bit, let me know if this is better.

Is the current wording fine?

Given the amount of changes from the original tender, I wonder if it makes sense to bounce this back to the ESC and discuss it in one of the next calls? We have the autoupdater tender discussion tomorrow already, so maybe it's a bit on too short notice and would fill the agenda too much, but it could be a topic in one of the next calls? Or is it better to suggest the above list via e-mail and use it if people are fine with it?

Let me know if discussing this in an ESC call works, or if you prefer to continue discussion on the development list, before we finalize the text here on board-discuss.

Thanks,
Florian

Hi Florian,

Regina (thanks a lot!) sent a list of bugs back in December on the dev
mailing list:
https://lists.freedesktop.org/archives/libreoffice/2021-December/088210.html

Was there any further discussion or feedback on this? If the list mentioned
there is fine, I replace item 2 from the tender with it. If we're unsure
whether that meets the budget or not, as the person days are listed in the
tender, we can add a note along the lines of "Please propose a subset and
prioritization of these bugs, that do not exceed the person days factored in
for this tender, see below."

I think this approach could work.

Regards,

Miklos

hi Florian,

Hello everyone,

the below mail is a bit older - Christmas break and some other tenders came in between, so I get to this only now.

(2)
The search result from item "Required 2." contains Meta-issues. Expanding them results in 80 issues.

Using Whiteboard as search criteria has no advantage compared to the Meta-issues. And I think both, Whitheboard search or Meta-issues, are not suitable for a tender, but a tender needs to list the issues explicitly.

The list from Whiteboard search and Meta-issues needs to be examined and prioritized manually.

This is taken from the specification at https://wiki.documentfoundation.org/Development/Budget2021#Cleanup_.26_further_improve_ODF_conformance

I fear answering that question is beyond my skills. :wink: Does it make sense to bounce this question back to the ESC for further specification?

Regina (thanks a lot!) sent a list of bugs back in December on the dev mailing list: https://lists.freedesktop.org/archives/libreoffice/2021-December/088210.html

Was there any further discussion or feedback on this? If the list mentioned there is fine, I replace item 2 from the tender with it. If we're unsure whether that meets the budget or not, as the person days are listed in the tender, we can add a note along the lines of "Please propose a subset and prioritization of these bugs, that do not exceed the person days factored in for this tender, see below."

thanks for reminding me, due to too much vacation i forgot but now i just provided some feedback about some of the issues to Regina.

i haven't thought about how much time the selected issues would require yet, it's possible it might still be more than the budget which the BoD wants to spend, so i guess a fixed budget and then apply for a subset of the issues makes sense.

the scope of this is quite large and unclear... *required* items are:

1. ODFAutoTests: addressing issues will be difficult because as Regina points out the web service appears to be offline.
   IIRC it's possible to run the tests offline, but currently i guess nobody knows how much work it is to set that up and what problems would actually be found, so i guess this item mostly amounts to "get ODFAutoTests to run at all".

I've tried to rephrase #1 a bit, let me know if this is better.

Is the current wording fine?

i guess, as long as nobody interprets it to mean "set up a public website" :slight_smile:

hello again,

(3)
Is it possible to get
https://bugs.documentfoundation.org/show_bug.cgi?id=48392
ODF: Implementation for svg:linearGradient and svg:radialGradient is missing
as explicit issue for "Required"?
We had this already as suggestion "Multi-color gradient" in
https://wiki.documentfoundation.org/Development/Budget2021
and now again in
https://wiki.documentfoundation.org/Development/Budget2022

I've added it. Not sure, however, how much that would change the work/cost estimate of the tender.

This is in the draft. We can add a similar note as mentioned above if we're unsure about the work required.

so i've discussed this with Armin now and we noticed that this would be a *lot* of effort and really should be a separate tender, and the gradients are currently listed separately in the Wiki page.

or, put another way, if you put the gradients into this tender there won't be any time to fix any other bug.

regards,
  michael

Hello,

thanks for your feedback, everyone!

I tried to factor in the changes and suggestions proposed into the tender draft at https://nextcloud.documentfoundation.org/s/ggqpciBK54rztJi, splitting the issues in high and low priority.

For Michael's suggestions, I wasn't sure about the priority, so put them into the low priority one, but likely that is wrong, so let me know what to put where.

Let me know also what you think of the approach in general.

Thanks a lot,
Florian