Discussion:
[edk2] Replacement EDK2 email list coming soon
Peterson, Joe
2015-05-04 20:21:38 UTC
Permalink
Hello all,

Due to community feedback, a new mailing list is being set up to replace this one. The new list will be hosted on Lists.01.org and should be more stable and consistent than this one.

The host has an opt-in policy and will not allow the current subscription list to be imported so you will need to subscribe yourself. The timing of the final conversion to the new list is still to be determined, but in the meantime you can sign up for the new list here: https://lists.01.org/mailman/listinfo/edk2

Please keep all relevant communications on this channel and do not use the new one for patches or questions yet. Feel free to post questions/comment/concerns to this current list.

Stay tuned for more updates... A list of the content changes / improvement being worked can be found here: http://www.tianocore.org/news/2015/05/01/UnderConst.html

Thank you.
Jordan Justen
2015-05-05 17:59:24 UTC
Permalink
Post by Peterson, Joe
Due to community feedback, a new mailing list is being set up to replace
this one. The new list will be hosted on Lists.01.org and should be more
stable and consistent than this one.
The host has an opt-in policy and will not allow the current subscription
list to be imported so you will need to subscribe yourself. The timing of
the final conversion to the new list is still to be determined, but in the
https://lists.01.org/mailman/listinfo/edk2
Why not edk2-devel?
Post by Peterson, Joe
Please keep all relevant communications on this channel and do not use the
new one for patches or questions yet. Feel free to post
questions/comment/concerns to this current list.
Stay tuned for more updates... A list of the content changes / improvement
http://www.tianocore.org/news/2015/05/01/UnderConst.html
!!

Ok, so let's move to github already! :)

-Jordan
Olivier Martin
2015-05-07 15:07:47 UTC
Permalink
Announcing Changes Coming Soon: "5. Open Discussions"

Let's hope in the future community feedbacks will be more taking in account.

I am happy to see this list of new changes but it would have been better to involve the community in the discussion for the chosen solutions.
As Jordan said Github already supports many of these features. But I have got the impression when I read 'Gerrit' and I can see the new mailing-list is based on Intel 01.org website that the new community will be based around an Intel home-made solution (I know Gerrit is not an Intel product but it needs a dedicated server to run).


-----Original Message-----
From: Jordan Justen [mailto:***@intel.com]
Sent: 05 May 2015 18:59
To: edk2-***@lists.sourceforge.net; Peterson, Joe
Subject: Re: [edk2] Replacement EDK2 email list coming soon
Post by Peterson, Joe
Due to community feedback, a new mailing list is being set up to replace
this one. The new list will be hosted on Lists.01.org and should be more
stable and consistent than this one.
The host has an opt-in policy and will not allow the current subscription
list to be imported so you will need to subscribe yourself. The timing of
the final conversion to the new list is still to be determined, but in the
https://lists.01.org/mailman/listinfo/edk2
Why not edk2-devel?
Post by Peterson, Joe
Please keep all relevant communications on this channel and do not use the
new one for patches or questions yet. Feel free to post
questions/comment/concerns to this current list.
Stay tuned for more updates... A list of the content changes / improvement
http://www.tianocore.org/news/2015/05/01/UnderConst.html
!!

Ok, so let's move to github already! :)

-Jordan

------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud Widest out-of-the-box monitoring support with 50+ applications Performance metrics, stats and reports that give you Actionable Insights Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
_______________________________________________
edk2-devel mailing list
edk2-***@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/edk2-devel


-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.

ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590
ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
B Cran
2015-05-07 15:20:02 UTC
Permalink
Post by Olivier Martin
Announcing Changes Coming Soon: "5. Open Discussions"
Let's hope in the future community feedbacks will be more taking in account.
Given that people have proposed solutions in recent months and nothing
changed, I suspect more of a 'benevolent dictator' approach may be needed
to improve things.
--
Bruce
Peterson, Joe
2015-05-13 17:01:52 UTC
Permalink
Sorry for the delay in responding to this. With regard to involving the community in a discussion of solutions, yes, there is no intention of making this an "Intel only" thing. The posting of the list shouldn't be seen as "this is what will be done," rather, this is the list we have developed based upon feedback from the community to date. If you have feedback or know of anything we missed, please provide your feedback here via the mailing list. Also, we do not intend to make this a home grown solution if we can avoid it.

Please be encouraged to post questions/comment/concerns to the mailing list.

Thanks,
-JEEP
Joe Peterson

-----Original Message-----
From: Olivier Martin [mailto:***@arm.com]
Sent: Thursday, May 07, 2015 8:08 AM
To: edk2-***@lists.sourceforge.net; Peterson, Joe
Subject: RE: [edk2] Replacement EDK2 email list coming soon

Announcing Changes Coming Soon: "5. Open Discussions"

Let's hope in the future community feedbacks will be more taking in account.

I am happy to see this list of new changes but it would have been better to involve the community in the discussion for the chosen solutions.
As Jordan said Github already supports many of these features. But I have got the impression when I read 'Gerrit' and I can see the new mailing-list is based on Intel 01.org website that the new community will be based around an Intel home-made solution (I know Gerrit is not an Intel product but it needs a dedicated server to run).


-----Original Message-----
From: Jordan Justen [mailto:***@intel.com]
Sent: 05 May 2015 18:59
To: edk2-***@lists.sourceforge.net; Peterson, Joe
Subject: Re: [edk2] Replacement EDK2 email list coming soon
Post by Peterson, Joe
Due to community feedback, a new mailing list is being set up to replace
this one. The new list will be hosted on Lists.01.org and should be more
stable and consistent than this one.
The host has an opt-in policy and will not allow the current subscription
list to be imported so you will need to subscribe yourself. The timing of
the final conversion to the new list is still to be determined, but in the
https://lists.01.org/mailman/listinfo/edk2
Why not edk2-devel?
Post by Peterson, Joe
Please keep all relevant communications on this channel and do not use the
new one for patches or questions yet. Feel free to post
questions/comment/concerns to this current list.
Stay tuned for more updates... A list of the content changes / improvement
http://www.tianocore.org/news/2015/05/01/UnderConst.html
!!

Ok, so let's move to github already! :)

-Jordan

------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud Widest out-of-the-box monitoring support with 50+ applications Performance metrics, stats and reports that give you Actionable Insights Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
_______________________________________________
edk2-devel mailing list
edk2-***@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/edk2-devel


-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.

ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
B Cran
2015-05-26 12:35:36 UTC
Permalink
Post by Peterson, Joe
Sorry for the delay in responding to this. With regard to involving the
community in a discussion of solutions, yes, there is no intention of
making this an "Intel only" thing. The posting of the list shouldn't be
seen as "this is what will be done," rather, this is the list we have
developed based upon feedback from the community to date. If you have
feedback or know of anything we missed, please provide your feedback here
via the mailing list. Also, we do not intend to make this a home grown
solution if we can avoid it.
Please be encouraged to post questions/comment/concerns to the mailing list.
I don't know how many others feel similarly, but I've seen very negative
feedback on Gerrit in the past at $work, mostly around its unfriendly and
ugly UI. If a code review system is going to be put into place, it might
make sense to use something that's more popular, in use by more open source
projects? Also, are there any plans to open up the tianocore.org site to
other contributors, so anyone (or at least active edk2-devel members) can
contribute and edit the pages?

Bruce
Laszlo Ersek
2015-05-26 13:56:18 UTC
Permalink
Post by Peterson, Joe
Sorry for the delay in responding to this. With regard to involving
the community in a discussion of solutions, yes, there is no
intention of making this an "Intel only" thing. The posting of the
list shouldn't be seen as "this is what will be done," rather, this
is the list we have developed based upon feedback from the community
to date. If you have feedback or know of anything we missed, please
provide your feedback here via the mailing list. Also, we do not
intend to make this a home grown solution if we can avoid it.
Please be encouraged to post questions/comment/concerns to the mailing list.
I don't know how many others feel similarly, but I've seen very negative
feedback on Gerrit in the past at $work, mostly around its unfriendly
and ugly UI. If a code review system is going to be put into place, it
might make sense to use something that's more popular, in use by more
open source projects?
Nothing comes close to reviews done in email (*). As I stated earlier
elsewhere:

- No web based application will ever be as flexible as free flowing text
for expressing thoughts about patches. I skimmed some upstream Gerrit
page / examples before, and what I saw could certainly not accommodate
the amount & format of text (eg. ASCII diagrams) that I
sometimes produce.

- Mailing lists are unbeaten at preserving threading, and at being
archived / mirrored without coordination.

(*) assuming (a) contributors use git-send-email to post patches, and
(b) reviewers use sensible MUAs that don't mangle plaintext,
monospace font emails.

Gerrit may be okay for tight-knit internal teams, but for
distributed development it's not appropriate in my opinion. Unless I'm
wrong, Gerrit has been designed for in-house development, from the
grounds up.

I don't intend to use Gerrit, and I very much hope that all contributors
will continue posting patches to, and accepting feedback from, the list.

Thanks
Laszlo
Post by Peterson, Joe
Also, are there any plans to open up the
tianocore.org <http://tianocore.org> site to other contributors, so
anyone (or at least active edk2-devel members) can contribute and edit
the pages?
Bruce
------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
Andrew Fish
2015-05-26 20:33:50 UTC
Permalink
Post by Laszlo Ersek
Post by Peterson, Joe
Sorry for the delay in responding to this. With regard to involving
the community in a discussion of solutions, yes, there is no
intention of making this an "Intel only" thing. The posting of the
list shouldn't be seen as "this is what will be done," rather, this
is the list we have developed based upon feedback from the community
to date. If you have feedback or know of anything we missed, please
provide your feedback here via the mailing list. Also, we do not
intend to make this a home grown solution if we can avoid it.
Please be encouraged to post questions/comment/concerns to the mailing list.
I don't know how many others feel similarly, but I've seen very negative
feedback on Gerrit in the past at $work, mostly around its unfriendly
and ugly UI. If a code review system is going to be put into place, it
might make sense to use something that's more popular, in use by more
open source projects?
Nothing comes close to reviews done in email (*). As I stated earlier
- No web based application will ever be as flexible as free flowing text
for expressing thoughts about patches. I skimmed some upstream Gerrit
page / examples before, and what I saw could certainly not accommodate
the amount & format of text (eg. ASCII diagrams) that I
sometimes produce.
- Mailing lists are unbeaten at preserving threading, and at being
archived / mirrored without coordination.
(*) assuming (a) contributors use git-send-email to post patches, and
(b) reviewers use sensible MUAs that don't mangle plaintext,
monospace font emails.
Gerrit may be okay for tight-knit internal teams, but for
distributed development it's not appropriate in my opinion. Unless I'm
wrong, Gerrit has been designed for in-house development, from the
grounds up.
I don't intend to use Gerrit, and I very much hope that all contributors
will continue posting patches to, and accepting feedback from, the list.
Is it possible to do both? So the Gerrit part is just a link that is auto added to the commit message?

I’ve only played with Gerrit one time, so I may be way off
. But this would be cool.

1) Submit patch to Gerrit, much like we do today to the mailing list.
2) Gerrit runs that patch against the server farm of known compilers and makes sure everything compiles. Kicks back the patch to the author on a build fail.
3) If everything compiles patch is sent to edk2 mailing list with link to Gerrit auto-added.

We can still required all the comments happen in the mailing list.
To me the win for Gerrit is workflow automation, getting to test compile against other tools. Also I’m a visual diff kind of person, so my brain likes seeing the visual diff of the change, vs a command line diff. It would be nice to have access to this on the web, vs. having to apply the patch to branch manually.

Thanks,

Andrew Fish
Post by Laszlo Ersek
Thanks
Laszlo
Post by Peterson, Joe
Also, are there any plans to open up the
tianocore.org <http://tianocore.org/> <http://tianocore.org <http://tianocore.org/>> site to other contributors, so
anyone (or at least active edk2-devel members) can contribute and edit
the pages?
Bruce
------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y <http://ad.doubleclick.net/ddm/clk/290420510;117567292;y>
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel <https://lists.sourceforge.net/lists/listinfo/edk2-devel>
Laszlo Ersek
2015-05-27 08:52:06 UTC
Permalink
Post by Andrew Fish
Post by Laszlo Ersek
On Wed, May 13, 2015 at 11:01 AM, Peterson, Joe
Sorry for the delay in responding to this. With regard to involving
the community in a discussion of solutions, yes, there is no
intention of making this an "Intel only" thing. The posting of the
list shouldn't be seen as "this is what will be done," rather, this
is the list we have developed based upon feedback from the community
to date. If you have feedback or know of anything we missed, please
provide your feedback here via the mailing list. Also, we do not
intend to make this a home grown solution if we can avoid it.
Please be encouraged to post questions/comment/concerns to the mailing list.
I don't know how many others feel similarly, but I've seen very negative
feedback on Gerrit in the past at $work, mostly around its unfriendly
and ugly UI. If a code review system is going to be put into place, it
might make sense to use something that's more popular, in use by more
open source projects?
Nothing comes close to reviews done in email (*). As I stated earlier
- No web based application will ever be as flexible as free flowing text
for expressing thoughts about patches. I skimmed some upstream Gerrit
page / examples before, and what I saw could certainly not accommodate
the amount & format of text (eg. ASCII diagrams) that I
sometimes produce.
- Mailing lists are unbeaten at preserving threading, and at being
archived / mirrored without coordination.
(*) assuming (a) contributors use git-send-email to post patches, and
(b) reviewers use sensible MUAs that don't mangle plaintext,
monospace font emails.
Gerrit may be okay for tight-knit internal teams, but for
distributed development it's not appropriate in my opinion. Unless I'm
wrong, Gerrit has been designed for in-house development, from the
grounds up.
I don't intend to use Gerrit, and I very much hope that all contributors
will continue posting patches to, and accepting feedback from, the list.
Is it possible to do both? So the Gerrit part is just a link that is
auto added to the commit message?
I’ve only played with Gerrit one time, so I may be way off…. But this
would be cool.
1) Submit patch to Gerrit, much like we do today to the mailing list.
2) Gerrit runs that patch against the server farm of known compilers and
makes sure everything compiles. Kicks back the patch to the author on a
build fail.
3) If everything compiles patch is sent to edk2 mailing list with link
to Gerrit auto-added.
We do something similar in the kernel and virt teams at RH. All patch
series (which are backports, most frequently) must be submitted to a
build farm first. The build must succeed for all supported platforms,
and (ideally) the developer should perform his / her final personal
testing with those binary packages that result from said build.

When everything's fine, the developer is allowed to post the series to
the appropriate internal list. All postings must carry (a) a Bugzilla
number, (b) a build identifier (a URL, practically). The build URL
sticks around forever -- packages are not preserved after a while, but
the metadata stays around forever.

Assuming the series gathers a sufficient number of ACKs (three on the
virt lists, variable / approx. three on the kernel lists), the
respective maintainers pick up the patches and apply them to the
internal repos.

... So, I definitely agree with the build farm idea; that would help
immensely in avoiding warnings-treated-as-errors that hit only on some
compilers. However, I'd prefer if the build URL / build ID were stuck
manually in the patch submission. I'd like to keep the integration /
automation between these two phases (central build & patch submission)
minimal. I'd like to keep retain control over the patch submission.

If there was very good tooling support (ie. git command line
integration) with gerrit, and the gerrit server was extremely stable,
then I might change my mind I guess. I've been using the process
described above for years; it matches / extends how big upstream
projects are developed (Linux kernel and QEMU for example) -- I trust
that workflow. On the other hand the gerrit sites / examples I've seen
thus far (from a distance, admittedly) looked awful, so I'm wary.
Post by Andrew Fish
We can still required all the comments happen in the mailing list.
To me the win for Gerrit is workflow automation, getting to test compile
against other tools.
I agree about the test compile 100%. I just wouldn't like to relinquish
control over the patch formatting and the submission. The mailing list
should remain the primary interface for contributions -- we must
accommodate drive-by contributors as well. Uploading patches to the
build farm and firing off test builds causes very real costs
(electricity, cooling, bandwidth) for the operator, so it would likely
be tied to authentication (just like RH's internal build farm, which is
similar to <http://koji.fedoraproject.org/>, is tied to authentication).
We shouldn't introduce barriers for newbies / one-off contributors for
*posting* their patches. For applying their patches, we can enforce
whatever we want; a seasoned developer could test-build patches for an
occasional contributor before applying & pushing his/her patches.
Post by Andrew Fish
Also I’m a visual diff kind of person, so my brain
likes seeing the visual diff of the change, vs a command line diff. It
would be nice to have access to this on the web, vs. having to apply the
patch to branch manually.
Access on the web looks like a step forward (eg. it provides syntax
highlighting), but it's actually a small step at a steep price. The
price is that a web browser (and a central server) are required.

A browser should not be a hard requirement at any point of the
submission and review process. (Test builds are a different matter,
although command line tools for those would be preferred as well.)

I wrote to Bruce earlier, in private, with regard to reviewing patches
Post by Andrew Fish
I agree. Which is why the reviewer should apply the patches from the
mailing list (or pull them from the submitter's public repo) and
review them with context.
Whenever I review a longer series, I create (or pull) a branch with
the patches, and as I progress through the review, I advance with the
local checkout as well. This way I have the full tree at my disposal,
in my own programmer's editor, and I can check even very distant
contexts against the patch, with the full power of my usual work
environment / desktop. I can grep, jump to tags, build in the middle
of the series, and so on.
There's no substitute for that.
And when I have things to say, I can insert them at the exact location
in the patch email; I can format or draw my comments the way I want...
So, patches, especially each patch in a larger series, should be
reviewed against a tree that has all *prior* patches in the series
applied -- ultimately that's how the contributor developed the patches
in the first place. In my opinion, a web based tool has no chance at
being as flexible at this (ie. at moving around in the entire tree
mid-series) as everyone's own desktop environment.

Assuming we agree on the above, let's discuss how people grab patches.
Normally, the easiest way to bring the patches from the mailing list to
a reviewer's desktop environment / local branch is (1) save the emails
from your MUA (a few clicks or keypresses), (2) apply them with "git am"
(a few more keypresses in your terminal). I agree that this presents
difficulties to many edk2 developers, because:

- They (have to?) use *gravely* sub-standard MUAs. Let's not name names,
but I'll say that a MUA that cannot get the absolute basics right
(comment at the bottom by default; keep threading pristine; only reflow
lines that end with a space etc) cannot be considered anything but
utterly broken. In theory such MUAs are plainly unsuitable for
distributed open source development; in practice we must cope with them.

- edk2 uses CRLF line terminators internally, and git-am has admittedly
problems with that, *if* the patches cross MTA boundaries (which is the
norm of course).

Therefore, we've grown to push edk2 series to personal github repos /
branches, and we reference them in the blurbs. Those branches can be
fetched very easily. (I do concede that with the same effort we might as
well push our branches to a gerrit server, if such pushing makes sense.)

... Okay, this email is again too long; sorry about that. Summary of my
opinion:

(1) The build farm is a great idea, and we should make its use a
requirement for *applying* someone's patches. (For example, Olivier
already does this with the ARM-internal continuous integration environment.)

(2) The barrier to *submit* patches should be low: the primary interface
should remain the mailing list. Pulling should be kept easy for
reviewers' (and testers') sake.

(3) A web browser should not be a required tool in performing the
actions inherent in submission and review. A browser and a web app might
be slightly more convenient for *consuming* short series than plaintext
email, but (a) for longer series, where individual patches need to be
reviewed against a continually advancing context (ie. tree), a web-based
tool is much weaker than a local development environment, providing
little benefit, (b) a web app will never be as flexible at *producing*
careful comments (think diagrams) as plain-text tools / MUAs.

(4) A web app is likely worse at preserving the threading of a
discussion than a mailing list. Also, data kept in a web app is less
suitable for independent archival / mirroring than a mailing list.

... In fact this makes me recall:

https://lwn.net/Articles/638090/ (article)
https://lwn.net/Articles/639051/ (comment by me)
https://lwn.net/Articles/639071/ (comment by me)

Thanks
Laszlo

------------------------------------------------------------------------------
Andrew Fish
2015-05-27 16:06:27 UTC
Permalink
Post by Laszlo Ersek
Post by Andrew Fish
Also I’m a visual diff kind of person, so my brain
likes seeing the visual diff of the change, vs a command line diff. It
would be nice to have access to this on the web, vs. having to apply the
patch to branch manually.
Access on the web looks like a step forward (eg. it provides syntax
highlighting), but it's actually a small step at a steep price. The
price is that a web browser (and a central server) are required.
“Stone Knives and bearskins” aside, why is a web browser bad?

The git repository is a centralized server, and so is the website.

Thanks,

Andrew Fish
Laszlo Ersek
2015-05-27 17:20:09 UTC
Permalink
Post by Laszlo Ersek
Post by Andrew Fish
Also I’m a visual diff kind of person, so my brain
likes seeing the visual diff of the change, vs a command line diff. It
would be nice to have access to this on the web, vs. having to apply the
patch to branch manually.
Access on the web looks like a step forward (eg. it provides syntax
highlighting), but it's actually a small step at a steep price. The
price is that a web browser (and a central server) are required.
“Stone Knives and bearskins” aside, why is a web browser bad?
Ultimately, I can only say that I've found web apps very limiting when
engaging in technical discussion. I find it hard to read code,
arguments, and diagrams in today's web apps, and I also find it very
difficult to present my ideas clearly. Web applications seem to optimize
for ease of (thoughtless) expression, and they encourage social
interaction rather than cultivating clarity of thought and structure.

Most people don't think in structure (even in technical discussions --
I'm witnessing it all over technical fora), so they might not perceive
the non-support for structure repressive. Personally I'm driven nuts by it.

* Here's an example: how could I write a review like

http://thread.gmane.org/gmane.comp.bios.tianocore.devel/3144/focus=3198

in a web application? (I'm referring to the tables and diagrams; please
keep scrolling down.) I'd have to fight with HTML tags or markup all the
time.

In addition, how could I reference that review two years later, like I
just did?

How could I search my local, cross-referenced archive for said review,
and find it in less than 60 seconds, like I just did? (I found the
Message-Id locally, and gmane allows remote lookup by Message-Id.)

* Another example: <https://bugs.launchpad.net/>. It's about the worst
bug tracker ever. It optimizes for "nice" over "useful". Evidence:

- It wants to prevent careless users from pasting text that is too wide,
thereby disrupting the appearance of the website. So it inserts <wbr>
tags into "overlong" words. Surprise, a full sized git commit hash
qualifies as overlong, so it breaks it into three pieces with
(invisible) <wbr> tags. The result is that double-click will not select
the full hash, only a third of it. Plus, the commit hash could be
wrapped if it was rendered near the end of a line, rather than forcing a
new line. How counterproductive is that?

- LP wants to prevent careless users from pasting overlong comments --
too many lines -- lest the appearance of the website be disrupted. So
they truncate each comment after a certain length. If you want to read
the full comment, you have to click a link, and then the comment is
shown in isolation. So you are forced to choose between "reading
elaborate comment in full" and "seeing elaborate comment in the context
of other comments". How counterproductive is that? For example, how can
you search the complete bug report (all comments at once) for a specific
string, using your browser's built-in "search in page" function? Well,
you can't.

* Another example: the Mailman 3.0 web archive. I referenced the LWN
article and my specific complaints earlier. It took a nose dive after 2.0.

* Another example: the ArchLinux forum / topic about GPU passthrough to
QEMU/KVM virtual machines, using VFIO, running OVMF + Windows, mainly.

https://bbs.archlinux.org/viewtopic.php?id=162768

The topic is over 5200 posts, and it is presented as 200+ pages of 25
posts each. How counterproductive is that? It is completely unusable.
The forum supports no threading (obviously), instead you see these box
quotes embedded in box quotes embedded in box quotes. There is no
support for searching, indexing, local mirroring. 5200 messages should
be trivial to search!

A standalone mailing list should have been created ages ago for this
topic. 5200 messages are a ridiculously low amount for a mailing list to
handle. Threading, mirroring, indexation, searching would come for free.

The subject matter is supposed to be fairly technical. I commented a few
times, only to grow a burning hate for the forum software in question.

My concern is not rooted in a penchant for "Stone Knives and bearskins".
I'm worried because all web apps seem to favor form over function, plus
accessibility via (extremely limiting) mobile devices over carefully
laid out arguments.

Participation via handheld devices, and eye candy, are worthless for
engineering discussions. The form of expression that they encourage
*restricts* thoughtful exchange.
The git repository is a centralized server, and so is the website.
The git repository supports "fetch commits" as a primary function. It
can be mirrored all over the world. Once you update your local clone,
the full power of your desktop (and the full project history) is at your
service. The git repo doesn't centralize control, or workflow.

I'm not sure what you mean by the website. If you mean the tianocore
wiki, or the current mailing list archive (on sourceforge), they are
indeed difficult to use. The wiki is hard to search (like most other
wikis), and the SF web archive is a joke (searching, threading... I'm
repeating myself).

Ultimately, I can only speak of personal experience. If yours differs, I
positively respect that. I just wouldn't like to see my productivity
hampered. :(

Thanks
Laszlo

------------------------------------------------------------------------------
Jordan Justen
2015-05-27 18:21:42 UTC
Permalink
Post by Laszlo Ersek
Post by Laszlo Ersek
Access on the web looks like a step forward (eg. it provides syntax
highlighting), but it's actually a small step at a steep price. The
price is that a web browser (and a central server) are required.
“Stone Knives and bearskins” aside, why is a web browser bad?
Ultimately, I can only say that I've found web apps very limiting when
engaging in technical discussion.
See also: this thread right here.

... Actual discussion happening. Can you imagine trying to do the same
on a web forum?

... Little barrier to entry to the discussion. (An email account.)

I happen to agree with Laszlo. We should retain the ability for people
to run git format-patch/send-email to contribute.

If we adopt some web based system, we should note that we are going to
cut some people out of that loop.

I've seen Gerrit a little, but I've not actually worked with it.
Regarding what I saw, I wasn't particularly impressed (nor concerned).
It didn't appear to bring all that much beyond email reviews.

-Jordan

------------------------------------------------------------------------------
Andrew Fish
2015-05-27 18:49:19 UTC
Permalink
Post by Jordan Justen
Post by Laszlo Ersek
Post by Laszlo Ersek
Access on the web looks like a step forward (eg. it provides syntax
highlighting), but it's actually a small step at a steep price. The
price is that a web browser (and a central server) are required.
“Stone Knives and bearskins” aside, why is a web browser bad?
Ultimately, I can only say that I've found web apps very limiting when
engaging in technical discussion.
See also: this thread right here.
... Actual discussion happening. Can you imagine trying to do the same
on a web forum?
... Little barrier to entry to the discussion. (An email account.)
I did not think a web browser was a barrier to entry, hence “Stone Knives and bearskins” Star Trek quote.

In my proposal I tried to position the web browser as adding value (visual diff for free, applies patch for free, auto builder for free), not being a required workflow.
Post by Jordan Justen
I happen to agree with Laszlo. We should retain the ability for people
to run git format-patch/send-email to contribute.
If we adopt some web based system, we should note that we are going to
cut some people out of that loop.
I've seen Gerrit a little, but I've not actually worked with it.
Regarding what I saw, I wasn't particularly impressed (nor concerned).
It didn't appear to bring all that much beyond email reviews.
I’m not tied to Gerrit. If we add an auto builder, then the review becomes more than send a patch to the mailing list. I’d be fine if that was an automated mailing list that did the auto build, if that passes post to the edk2 mailing list and maybe have some quick way to do a graphical diff. Maybe that is not possible.

I think what we are saying is a new way to review patches is not worth changing the current workflow. If we got an auto builder, or for me a quick way to look at patches with out having to apply them, then maybe changing the work flow is worth it. I’m fine with doing all the feedback on the mailing list.

Thanks,

Andrew Fish
Post by Jordan Justen
-Jordan
------------------------------------------------------------------------------
Carsey, Jaben
2015-05-27 18:55:54 UTC
Permalink
Post by Olivier Martin
-----Original Message-----
Sent: Wednesday, May 27, 2015 11:49 AM
To: Justen, Jordan L
Subject: Re: [edk2] Replacement EDK2 email list coming soon
Post by Jordan Justen
Post by Laszlo Ersek
Post by Laszlo Ersek
Access on the web looks like a step forward (eg. it provides syntax
highlighting), but it's actually a small step at a steep price. The
price is that a web browser (and a central server) are required.
“Stone Knives and bearskins” aside, why is a web browser bad?
Ultimately, I can only say that I've found web apps very limiting
when engaging in technical discussion.
See also: this thread right here.
... Actual discussion happening. Can you imagine trying to do the same
on a web forum?
... Little barrier to entry to the discussion. (An email account.)
I did not think a web browser was a barrier to entry, hence “Stone Knives and
bearskins” Star Trek quote.
In my proposal I tried to position the web browser as adding value (visual diff
for free, applies patch for free, auto builder for free), not being a required
workflow.
Post by Jordan Justen
I happen to agree with Laszlo. We should retain the ability for people
to run git format-patch/send-email to contribute.
If we adopt some web based system, we should note that we are going to
cut some people out of that loop.
I've seen Gerrit a little, but I've not actually worked with it.
Regarding what I saw, I wasn't particularly impressed (nor concerned).
It didn't appear to bring all that much beyond email reviews.
I’m not tied to Gerrit. If we add an auto builder, then the review becomes
more than send a patch to the mailing list. I’d be fine if that was an
automated mailing list that did the auto build, if that passes post to the edk2
mailing list and maybe have some quick way to do a graphical diff. Maybe
that is not possible.
Graphical diff is frequently much more useful than the email. Often I need more surrounding code than the email has.
Post by Olivier Martin
I think what we are saying is a new way to review patches is not worth
changing the current workflow. If we got an auto builder, or for me a quick
way to look at patches with out having to apply them, then maybe changing
the work flow is worth it. I’m fine with doing all the feedback on the mailing
list.
Email based feedback works well so far as I can tell.
Post by Olivier Martin
Thanks,
Andrew Fish
Post by Jordan Justen
-Jordan
------------------------------------------------------------------------------
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
Laszlo Ersek
2015-05-27 19:06:21 UTC
Permalink
Post by Carsey, Jaben
Post by Olivier Martin
-----Original Message-----
Sent: Wednesday, May 27, 2015 11:49 AM
To: Justen, Jordan L
Subject: Re: [edk2] Replacement EDK2 email list coming soon
Post by Jordan Justen
Post by Laszlo Ersek
Post by Laszlo Ersek
Access on the web looks like a step forward (eg. it provides syntax
highlighting), but it's actually a small step at a steep price. The
price is that a web browser (and a central server) are required.
“Stone Knives and bearskins” aside, why is a web browser bad?
Ultimately, I can only say that I've found web apps very limiting
when engaging in technical discussion.
See also: this thread right here.
... Actual discussion happening. Can you imagine trying to do the same
on a web forum?
... Little barrier to entry to the discussion. (An email account.)
I did not think a web browser was a barrier to entry, hence “Stone Knives and
bearskins” Star Trek quote.
In my proposal I tried to position the web browser as adding value (visual diff
for free, applies patch for free, auto builder for free), not being a required
workflow.
Post by Jordan Justen
I happen to agree with Laszlo. We should retain the ability for people
to run git format-patch/send-email to contribute.
If we adopt some web based system, we should note that we are going to
cut some people out of that loop.
I've seen Gerrit a little, but I've not actually worked with it.
Regarding what I saw, I wasn't particularly impressed (nor concerned).
It didn't appear to bring all that much beyond email reviews.
I’m not tied to Gerrit. If we add an auto builder, then the review becomes
more than send a patch to the mailing list. I’d be fine if that was an
automated mailing list that did the auto build, if that passes post to the edk2
mailing list and maybe have some quick way to do a graphical diff. Maybe
that is not possible.
Graphical diff is frequently much more useful than the email. Often I need more surrounding code than the email has.
I agree 100% (as I explained. Probably with too many words.)

The solution is, IMHO, not to move to a web app, but to make it very
easy for reviewers to apply / fetch the patch series onto / into their
local clones. This way the entire *tree* is at the reviewer's disposal,
at any stage in the series, and not just for review, but for building
and testing as well. Diffs can be displayed with everyone's favorite tool.

More context than that simply doesn't exist.

Thanks
Laszlo
Post by Carsey, Jaben
Post by Olivier Martin
I think what we are saying is a new way to review patches is not worth
changing the current workflow. If we got an auto builder, or for me a quick
way to look at patches with out having to apply them, then maybe changing
the work flow is worth it. I’m fine with doing all the feedback on the mailing
list.
Email based feedback works well so far as I can tell.
Post by Olivier Martin
Thanks,
Andrew Fish
Post by Jordan Justen
-Jordan
------------------------------------------------------------------------------
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
Ard Biesheuvel
2015-06-03 09:12:29 UTC
Permalink
Post by Laszlo Ersek
Post by Carsey, Jaben
Post by Olivier Martin
-----Original Message-----
Sent: Wednesday, May 27, 2015 11:49 AM
To: Justen, Jordan L
Subject: Re: [edk2] Replacement EDK2 email list coming soon
Post by Jordan Justen
Post by Laszlo Ersek
Post by Laszlo Ersek
Access on the web looks like a step forward (eg. it provides syntax
highlighting), but it's actually a small step at a steep price. The
price is that a web browser (and a central server) are required.
“Stone Knives and bearskins” aside, why is a web browser bad?
Ultimately, I can only say that I've found web apps very limiting
when engaging in technical discussion.
See also: this thread right here.
... Actual discussion happening. Can you imagine trying to do the same
on a web forum?
... Little barrier to entry to the discussion. (An email account.)
I did not think a web browser was a barrier to entry, hence “Stone Knives and
bearskins” Star Trek quote.
In my proposal I tried to position the web browser as adding value (visual diff
for free, applies patch for free, auto builder for free), not being a required
workflow.
Post by Jordan Justen
I happen to agree with Laszlo. We should retain the ability for people
to run git format-patch/send-email to contribute.
If we adopt some web based system, we should note that we are going to
cut some people out of that loop.
I've seen Gerrit a little, but I've not actually worked with it.
Regarding what I saw, I wasn't particularly impressed (nor concerned).
It didn't appear to bring all that much beyond email reviews.
I’m not tied to Gerrit. If we add an auto builder, then the review becomes
more than send a patch to the mailing list. I’d be fine if that was an
automated mailing list that did the auto build, if that passes post to the edk2
mailing list and maybe have some quick way to do a graphical diff. Maybe
that is not possible.
Graphical diff is frequently much more useful than the email. Often I need more surrounding code than the email has.
I agree 100% (as I explained. Probably with too many words.)
The solution is, IMHO, not to move to a web app, but to make it very
easy for reviewers to apply / fetch the patch series onto / into their
local clones. This way the entire *tree* is at the reviewer's disposal,
at any stage in the series, and not just for review, but for building
and testing as well. Diffs can be displayed with everyone's favorite tool.
More context than that simply doesn't exist.
Something like patchwork/pwclient perhaps?

http://jk.ozlabs.org/projects/patchwork/

It is currently being used by a number of Linux kernel subsystem
maintainers, but it is entirely optional for contributors: it is
basically just a custom view onto the mailing list feed, with some
additional metadata to keep track of the acceptance state of patches.
Post by Laszlo Ersek
Post by Carsey, Jaben
Post by Olivier Martin
I think what we are saying is a new way to review patches is not worth
changing the current workflow. If we got an auto builder, or for me a quick
way to look at patches with out having to apply them, then maybe changing
the work flow is worth it. I’m fine with doing all the feedback on the mailing
list.
Email based feedback works well so far as I can tell.
Post by Olivier Martin
Thanks,
Andrew Fish
Post by Jordan Justen
-Jordan
------------------------------------------------------------------------------
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
Laszlo Ersek
2015-06-03 11:17:21 UTC
Permalink
Post by Ard Biesheuvel
Post by Laszlo Ersek
Post by Carsey, Jaben
Post by Olivier Martin
-----Original Message-----
Sent: Wednesday, May 27, 2015 11:49 AM
To: Justen, Jordan L
Subject: Re: [edk2] Replacement EDK2 email list coming soon
Post by Jordan Justen
Post by Laszlo Ersek
Post by Laszlo Ersek
Access on the web looks like a step forward (eg. it provides syntax
highlighting), but it's actually a small step at a steep price. The
price is that a web browser (and a central server) are required.
“Stone Knives and bearskins” aside, why is a web browser bad?
Ultimately, I can only say that I've found web apps very limiting
when engaging in technical discussion.
See also: this thread right here.
... Actual discussion happening. Can you imagine trying to do the same
on a web forum?
... Little barrier to entry to the discussion. (An email account.)
I did not think a web browser was a barrier to entry, hence “Stone Knives and
bearskins” Star Trek quote.
In my proposal I tried to position the web browser as adding value (visual diff
for free, applies patch for free, auto builder for free), not being a required
workflow.
Post by Jordan Justen
I happen to agree with Laszlo. We should retain the ability for people
to run git format-patch/send-email to contribute.
If we adopt some web based system, we should note that we are going to
cut some people out of that loop.
I've seen Gerrit a little, but I've not actually worked with it.
Regarding what I saw, I wasn't particularly impressed (nor concerned).
It didn't appear to bring all that much beyond email reviews.
I’m not tied to Gerrit. If we add an auto builder, then the review becomes
more than send a patch to the mailing list. I’d be fine if that was an
automated mailing list that did the auto build, if that passes post to the edk2
mailing list and maybe have some quick way to do a graphical diff. Maybe
that is not possible.
Graphical diff is frequently much more useful than the email. Often I need more surrounding code than the email has.
I agree 100% (as I explained. Probably with too many words.)
The solution is, IMHO, not to move to a web app, but to make it very
easy for reviewers to apply / fetch the patch series onto / into their
local clones. This way the entire *tree* is at the reviewer's disposal,
at any stage in the series, and not just for review, but for building
and testing as well. Diffs can be displayed with everyone's favorite tool.
More context than that simply doesn't exist.
Something like patchwork/pwclient perhaps?
http://jk.ozlabs.org/projects/patchwork/
It is currently being used by a number of Linux kernel subsystem
maintainers, but it is entirely optional for contributors: it is
basically just a custom view onto the mailing list feed, with some
additional metadata to keep track of the acceptance state of patches.
I consider patchwork
(a) primarily a maintainer tool (for tracking pending patches, tags from
others etc),
(b) secondarily a reviewer coordination tool -- sometimes people are
looking for stuff to review, and giving them subject lines and (more
importantly) Message-Ids is effective. (The review still happens on the
list.)

Laszlo
Post by Ard Biesheuvel
Post by Laszlo Ersek
Post by Carsey, Jaben
Post by Olivier Martin
I think what we are saying is a new way to review patches is not worth
changing the current workflow. If we got an auto builder, or for me a quick
way to look at patches with out having to apply them, then maybe changing
the work flow is worth it. I’m fine with doing all the feedback on the mailing
list.
Email based feedback works well so far as I can tell.
Post by Olivier Martin
Thanks,
Andrew Fish
Post by Jordan Justen
-Jordan
------------------------------------------------------------------------------
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
_______________________________________________
edk2-devel mailing list
https://lists.sourceforge.net/lists/listinfo/edk2-devel
------------------------------------------------------------------------------
Bruce Cran
2015-05-27 20:07:24 UTC
Permalink
Graphical diff is frequently much more useful than the email. Often I
need more surrounding code than the email has.
[Sending again, since my original reply didn't go to the ML.]

Would it be useful if I was to set up a test server for people to use
for a while to see if it would be useful, or add value?
--
Bruce
B Cran
2015-07-04 22:56:05 UTC
Permalink
Due to community feedback, a new mailing list is being set up to replace this one. The new list will be hosted on Lists.01.org <http://lists.01.org/> and should be more stable and consistent than this one.
The host has an opt-in policy and will not allow the current subscription list to be imported so you will need to subscribe yourself. The timing of the final conversion to the new list is still to be determined, but in the meantime you can sign up for the new list here: https://lists.01.org/mailman/listinfo/edk2 <https://lists.01.org/mailman/listinfo/edk2>
Please keep all relevant communications on this channel and do not use the new one for patches or questions yet. Feel free to post questions/comment/concerns to this current list.
Stay tuned for more updates
 A list of the content changes / improvement being worked can be found here:http:// <http://www.tianocore.org/news/2015/05/01/UnderConst.html>www.tianocore.org/news/2015/05/01/UnderConst.html <http://www.tianocore.org/news/2015/05/01/UnderConst.html>
Since it has been about 2 months since this was sent, I was wondering if there’s any update regarding plans, timing etc.?

—
Bruce Cran
Jordan Justen
2015-07-06 18:48:48 UTC
Permalink
Post by B Cran
Post by Peterson, Joe
Due to community feedback, a new mailing list is being set up to replace
this one. The new list will be hosted on Lists.01.org and should be more
stable and consistent than this one.
The host has an opt-in policy and will not allow the current
subscription list to be imported so you will need to subscribe yourself.
The timing of the final conversion to the new list is still to be
determined, but in the meantime you can sign up for the new list
here: https://lists.01.org/mailman/listinfo/edk2
Please keep all relevant communications on this channel and do not use
the new one for patches or questions yet. Feel free to post
questions/comment/concerns to this current list.
Stay tuned for more updates... A list of the content changes /
improvement being worked can be found
here:http://www.tianocore.org/news/2015/05/01/UnderConst.html
Since it has been about 2 months since this was sent, I was wondering if
there's any update regarding plans, timing etc.?
Joe is out of the office currently, so I'm going to take over on this
task. I plan to transition the list over the next week or two. (I'll
send out a separate email with the details.)

Note that compared to the original announcement the list has been
renamed from edk2 to edk2-devel.

https://lists.01.org/mailman/listinfo/edk2-devel

-Jordan

Loading...