sim: replacing ChangeLog files with online git logs

sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

we're planning on dropping ChangeLog files in sim/ and replacing
them with a pointer to the online vcs log.  anyone have feedback
before we land this ?

$ find sim/ -name ChangeLog -delete
$ cat >sim/ChangeLog <<EOF
Please see the git log online at:
  https://sourceware.org/git/?p=binutils-gdb.git;a=history;f=sim
EOF
-mike

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Date: Fri, 12 Mar 2021 21:53:54 -0500
> From: Mike Frysinger via Gdb <gdb@sourceware.org>
> 
> we're planning on dropping ChangeLog files in sim/ and replacing
> them with a pointer to the online vcs log.

Will the release tarballs include the generated ChangeLog file?  If
not, why not?

Thanks.

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 13 Mar 2021 09:30, Eli Zaretskii wrote:
> > Date: Fri, 12 Mar 2021 21:53:54 -0500
> > From: Mike Frysinger via Gdb <gdb@sourceware.org>
> > 
> > we're planning on dropping ChangeLog files in sim/ and replacing
> > them with a pointer to the online vcs log.
> 
> Will the release tarballs include the generated ChangeLog file?  If
> not, why not?

we can, but we're leaning towards not because they don't seem to be
providing any actual value for people.
-mike

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Date: Sat, 13 Mar 2021 10:12:40 -0500
> From: Mike Frysinger <vapier@gentoo.org>
> Cc: gdb@sourceware.org, andrew.burgess@embecosm.com
> 
> On 13 Mar 2021 09:30, Eli Zaretskii wrote:
> > > Date: Fri, 12 Mar 2021 21:53:54 -0500
> > > From: Mike Frysinger via Gdb <gdb@sourceware.org>
> > > 
> > > we're planning on dropping ChangeLog files in sim/ and replacing
> > > them with a pointer to the online vcs log.
> > 
> > Will the release tarballs include the generated ChangeLog file?  If
> > not, why not?
> 
> we can, but we're leaning towards not because they don't seem to be
> providing any actual value for people.

So people who don't have Git will be left without the logs?  I'm not
sure I'm okay with that.

Producing a ChangeLog from Git log when the release is tarred is not
hard, why skip it?

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 13 Mar 2021 18:02, Eli Zaretskii wrote:
> > Date: Sat, 13 Mar 2021 10:12:40 -0500
> > From: Mike Frysinger <vapier@gentoo.org>
> > Cc: gdb@sourceware.org, andrew.burgess@embecosm.com
> > 
> > On 13 Mar 2021 09:30, Eli Zaretskii wrote:
> > > > Date: Fri, 12 Mar 2021 21:53:54 -0500
> > > > From: Mike Frysinger via Gdb <gdb@sourceware.org>
> > > > 
> > > > we're planning on dropping ChangeLog files in sim/ and replacing
> > > > them with a pointer to the online vcs log.
> > > 
> > > Will the release tarballs include the generated ChangeLog file?  If
> > > not, why not?
> > 
> > we can, but we're leaning towards not because they don't seem to be
> > providing any actual value for people.
> 
> So people who don't have Git will be left without the logs?  I'm not
> sure I'm okay with that.

what value are people actually pulling from these ?  we aren't talking
about user-focused NEWS, we're talking about developer-focused ChangeLog.

> Producing a ChangeLog from Git log when the release is tarred is not
> hard, why skip it?

the hard & bikeshed part is what form does the log take.  do you just want
the default `git log` output ?
-mike

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Date: Sat, 13 Mar 2021 11:26:11 -0500
> From: Mike Frysinger <vapier@gentoo.org>
> Cc: gdb@sourceware.org, andrew.burgess@embecosm.com
> 
> > So people who don't have Git will be left without the logs?  I'm not
> > sure I'm okay with that.
> 
> what value are people actually pulling from these ?  we aren't talking
> about user-focused NEWS, we're talking about developer-focused ChangeLog.

This has been beaten to death in various forums, including GNU
internal ones.  the conclusion was that some people want that, and
some developers are not always on-line and with Git repo handy.

> the hard & bikeshed part is what form does the log take.  do you just want
> the default `git log` output ?

There's a Gnulib script which will produce the form that matches the
ChangeLog format.

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 13 Mar 2021 18:38, Eli Zaretskii wrote:
> > Date: Sat, 13 Mar 2021 11:26:11 -0500
> > From: Mike Frysinger <vapier@gentoo.org>
> > Cc: gdb@sourceware.org, andrew.burgess@embecosm.com
> > 
> > > So people who don't have Git will be left without the logs?  I'm not
> > > sure I'm okay with that.
> > 
> > what value are people actually pulling from these ?  we aren't talking
> > about user-focused NEWS, we're talking about developer-focused ChangeLog.
> 
> This has been beaten to death in various forums, including GNU
> internal ones.  the conclusion was that some people want that, and
> some developers are not always on-line and with Git repo handy.

the conclusion is that GNU standards made recommendations, but individual
projects can pick between them.

> > the hard & bikeshed part is what form does the log take.  do you just want
> > the default `git log` output ?
> 
> There's a Gnulib script which will produce the form that matches the
> ChangeLog format.

there's actually multiple gnulib scripts.  which one are you referring to ?
gitlog-to-changelog ?
-mike

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Date: Sat, 13 Mar 2021 13:21:29 -0500
> From: Mike Frysinger <vapier@gentoo.org>
> Cc: gdb@sourceware.org, andrew.burgess@embecosm.com
> 
> On 13 Mar 2021 18:38, Eli Zaretskii wrote:
> > > Date: Sat, 13 Mar 2021 11:26:11 -0500
> > > From: Mike Frysinger <vapier@gentoo.org>
> > > Cc: gdb@sourceware.org, andrew.burgess@embecosm.com
> > > 
> > > > So people who don't have Git will be left without the logs?  I'm not
> > > > sure I'm okay with that.
> > > 
> > > what value are people actually pulling from these ?  we aren't talking
> > > about user-focused NEWS, we're talking about developer-focused ChangeLog.
> > 
> > This has been beaten to death in various forums, including GNU
> > internal ones.  the conclusion was that some people want that, and
> > some developers are not always on-line and with Git repo handy.
> 
> the conclusion is that GNU standards made recommendations, but individual
> projects can pick between them.

That's a different conclusions.  You asked what people are doing with
the ChangeLogs, not whether it was mandatory to keep them.

> > There's a Gnulib script which will produce the form that matches the
> > ChangeLog format.
> 
> there's actually multiple gnulib scripts.  which one are you referring to ?
> gitlog-to-changelog ?

Yes, that one.

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 13 Mar 2021 20:32, Eli Zaretskii wrote:
> From: Mike Frysinger <vapier@gentoo.org>
> > On 13 Mar 2021 18:38, Eli Zaretskii wrote:
> > > From: Mike Frysinger <vapier@gentoo.org>
> > > > > So people who don't have Git will be left without the logs?  I'm not
> > > > > sure I'm okay with that.
> > > > 
> > > > what value are people actually pulling from these ?  we aren't talking
> > > > about user-focused NEWS, we're talking about developer-focused ChangeLog.
> > > 
> > > This has been beaten to death in various forums, including GNU
> > > internal ones.  the conclusion was that some people want that, and
> > > some developers are not always on-line and with Git repo handy.
> > 
> > the conclusion is that GNU standards made recommendations, but individual
> > projects can pick between them.
> 
> That's a different conclusions.  You asked what people are doing with
> the ChangeLogs, not whether it was mandatory to keep them.

that's fair.  my point/goal was that if there aren't requirements for it,
and there isn't a compelling use case for keeping them, then the easiest
thing is to do nothing and delete them.

it seems that there isn't a requirement ->
i'm skeptical that the logs add value as anyone who is going to contribute
by diving into the source history would be online w/git access ->
we punt the logs and stop wasting lots of developer time.
-mike

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Date: Wed, 17 Mar 2021 02:07:46 -0400
> From: Mike Frysinger <vapier@gentoo.org>
> Cc: gdb@sourceware.org, andrew.burgess@embecosm.com
> 
> > That's a different conclusions.  You asked what people are doing with
> > the ChangeLogs, not whether it was mandatory to keep them.
> 
> that's fair.  my point/goal was that if there aren't requirements for it,
> and there isn't a compelling use case for keeping them, then the easiest
> thing is to do nothing and delete them.
> 
> it seems that there isn't a requirement ->

Indeed, there's no mandatory requirement to keep the files, not
anymore.

> i'm skeptical that the logs add value as anyone who is going to contribute
> by diving into the source history would be online w/git access ->
> we punt the logs and stop wasting lots of developer time.

I'm not sure I agree with your skepticism.  And OTOH, producing the
ChangeLog from Git logs takes mere seconds and wastes no development
time.  So I think on balance there's very little price to pay for
something that some people may value.  Why refuse to pay that small
price and humor those few people?

Re: sim: replacing ChangeLog files with online git logs

[Luis Machado]

On 3/17/21 10:14 AM, Eli Zaretskii via Gdb wrote:
>> Date: Wed, 17 Mar 2021 02:07:46 -0400
>> From: Mike Frysinger <vapier@gentoo.org>
>> Cc: gdb@sourceware.org, andrew.burgess@embecosm.com
>>
>>> That's a different conclusions.  You asked what people are doing with
>>> the ChangeLogs, not whether it was mandatory to keep them.
>>
>> that's fair.  my point/goal was that if there aren't requirements for it,
>> and there isn't a compelling use case for keeping them, then the easiest
>> thing is to do nothing and delete them.
>>
>> it seems that there isn't a requirement ->
> 
> Indeed, there's no mandatory requirement to keep the files, not
> anymore.
> 
>> i'm skeptical that the logs add value as anyone who is going to contribute
>> by diving into the source history would be online w/git access ->
>> we punt the logs and stop wasting lots of developer time.
> 
> I'm not sure I agree with your skepticism.  And OTOH, producing the
> ChangeLog from Git logs takes mere seconds and wastes no development
> time.  So I think on balance there's very little price to pay for
> something that some people may value.  Why refuse to pay that small
> price and humor those few people?
> 

Are you proposing simply feeding the git log through the script and 
saving that to a file? Even if the output is less detailed than the 
ChangeLog entries we have now?

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Cc: gdb@sourceware.org
> From: Luis Machado <luis.machado@linaro.org>
> Date: Wed, 17 Mar 2021 11:31:25 -0300
> 
> > I'm not sure I agree with your skepticism.  And OTOH, producing the
> > ChangeLog from Git logs takes mere seconds and wastes no development
> > time.  So I think on balance there's very little price to pay for
> > something that some people may value.  Why refuse to pay that small
> > price and humor those few people?
> 
> Are you proposing simply feeding the git log through the script and 
> saving that to a file? Even if the output is less detailed than the 
> ChangeLog entries we have now?

That's the idea, yes.  Of course, for such a generated ChangeLog to be
useful, the Git commit log messages should be informative enough.

Re: sim: replacing ChangeLog files with online git logs

[Luis Machado]

On 3/17/21 11:47 AM, Eli Zaretskii wrote:
>> Cc: gdb@sourceware.org
>> From: Luis Machado <luis.machado@linaro.org>
>> Date: Wed, 17 Mar 2021 11:31:25 -0300
>>
>>> I'm not sure I agree with your skepticism.  And OTOH, producing the
>>> ChangeLog from Git logs takes mere seconds and wastes no development
>>> time.  So I think on balance there's very little price to pay for
>>> something that some people may value.  Why refuse to pay that small
>>> price and humor those few people?
>>
>> Are you proposing simply feeding the git log through the script and
>> saving that to a file? Even if the output is less detailed than the
>> ChangeLog entries we have now?
> 
> That's the idea, yes.  Of course, for such a generated ChangeLog to be
> useful, the Git commit log messages should be informative enough.
> 

I see. Given the ChangeLog will list all the files that were changed, I 
suppose the commit messages would only need to describe what the changes 
are about. But the description may not list function names, files, new 
data structures etc. Would that work?

Or did you have something more specific in mind?

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Cc: vapier@gentoo.org, gdb@sourceware.org
> From: Luis Machado <luis.machado@linaro.org>
> Date: Wed, 17 Mar 2021 11:52:03 -0300
> 
> > That's the idea, yes.  Of course, for such a generated ChangeLog to be
> > useful, the Git commit log messages should be informative enough.
> 
> I see. Given the ChangeLog will list all the files that were changed, I 
> suppose the commit messages would only need to describe what the changes 
> are about. But the description may not list function names, files, new 
> data structures etc. Would that work?
> 
> Or did you have something more specific in mind?

Ideally, the commit log messages should provide the same information
as the manually managed ChangeLog files, in which case generating the
files from Git log at release time doesn't lose any information.  If
some of that information is missing from the Git logs, then the
generated ChangeLog files will be somewhat less useful.  But if the
project doesn't require to record more than that in Git log messages,
then it's the problem with the Git logs, not with ChangeLogs.

For example, in Emacs we ask contributors to format the Git log
messages according to ChangeLog format, and then the generated
ChangeLog file looks exactly like the manual one would.

Re: sim: replacing ChangeLog files with online git logs

[Luis Machado]

On 3/17/21 12:09 PM, Eli Zaretskii wrote:
>> Cc: vapier@gentoo.org, gdb@sourceware.org
>> From: Luis Machado <luis.machado@linaro.org>
>> Date: Wed, 17 Mar 2021 11:52:03 -0300
>>
>>> That's the idea, yes.  Of course, for such a generated ChangeLog to be
>>> useful, the Git commit log messages should be informative enough.
>>
>> I see. Given the ChangeLog will list all the files that were changed, I
>> suppose the commit messages would only need to describe what the changes
>> are about. But the description may not list function names, files, new
>> data structures etc. Would that work?
>>
>> Or did you have something more specific in mind?
> 
> Ideally, the commit log messages should provide the same information
> as the manually managed ChangeLog files, in which case generating the
> files from Git log at release time doesn't lose any information.  If
> some of that information is missing from the Git logs, then the
> generated ChangeLog files will be somewhat less useful.  But if the
> project doesn't require to record more than that in Git log messages,
> then it's the problem with the Git logs, not with ChangeLogs.

The commit message, at least for the GDB project, includes a subject 
line, the description of the changes and an embedded ChangeLog entry. So 
there are 3 sections. sim/ seems to follow the same strategy.

Given the idea is to drop ChangeLog files for sim/ (for reasons Mike has 
exposed), it would make no sense to keep embedding a manually-written 
ChangeLog entry in the git commit message. So the only two sections left 
in the commit message would be the subject line and the description.

The list of changes would be auto-generated by a script, but it will 
certainly be less descriptive than a manually-written ChangeLog entry. 
It will only list things like "modified" or "new file".

> 
> For example, in Emacs we ask contributors to format the Git log
> messages according to ChangeLog format, and then the generated
> ChangeLog file looks exactly like the manual one would.
> 

I may be mistaken, but I think the idea is to drop ChangeLog 
dependencies even from commit messages. That will simplify the 
contribution process given ChangeLog files/entries will be 
auto-generated by the mentioned script.

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Cc: vapier@gentoo.org, gdb@sourceware.org
> From: Luis Machado <luis.machado@linaro.org>
> Date: Wed, 17 Mar 2021 12:22:46 -0300
> 
> Given the idea is to drop ChangeLog files for sim/ (for reasons Mike has 
> exposed), it would make no sense to keep embedding a manually-written 
> ChangeLog entry in the git commit message. So the only two sections left 
> in the commit message would be the subject line and the description.

That doesn't have to be so, though.  How to format the Git log
messages when ChangeLog files are dropped is a separate discussion.

> The list of changes would be auto-generated by a script, but it will 
> certainly be less descriptive than a manually-written ChangeLog entry. 
> It will only list things like "modified" or "new file".
> 
> > 
> > For example, in Emacs we ask contributors to format the Git log
> > messages according to ChangeLog format, and then the generated
> > ChangeLog file looks exactly like the manual one would.
> > 
> 
> I may be mistaken, but I think the idea is to drop ChangeLog 
> dependencies even from commit messages. That will simplify the 
> contribution process given ChangeLog files/entries will be 
> auto-generated by the mentioned script.

The list of files touched by a changeset can be easily generated, but
the functions/macros/classes not so easily, and not necessarily
accurately enough, even if we will use the more elaborate script that
is used by glibc.  So some information loss will be present.

OTOH, the ChangeLog-format Git log messages don't require one to have
the ChangeLog file, they just require to use the same format when
describing the changeset.  If you use Emacs to make the changes
themselves and commit the changes into Git, you just need to use the
Emacs commands to format the Git log messages, those commands will
produce the same information as the manual ChangeLog files
automatically.

If that is not an option that this community wants to adopt, then
gitlog-to-changelog is probably not the adequate solution, and we
should try using vcs-to-changelog.py (also in Gnulib).

Re: sim: replacing ChangeLog files with online git logs

[Simon Marchi]

We're getting into this again, I know you've said the horse has been
beaten to death about ChangeLogs, but it keeps coming up because it's a
real daily pain for many people.

On 2021-03-17 11:40 a.m., Eli Zaretskii via Gdb wrote:
> The list of files touched by a changeset can be easily generated, but
> the functions/macros/classes not so easily, and not necessarily
> accurately enough, even if we will use the more elaborate script that
> is used by glibc.  So some information loss will be present.
> 
> OTOH, the ChangeLog-format Git log messages don't require one to have
> the ChangeLog file, they just require to use the same format when
> describing the changeset.

It's not the ChangeLog file per-se that's annoying, it's writing the
entry.  It's tedious and contains mostly uninteresting information.  So
putting the ChangeLog entry in the git log instead of a file doesn't
change anything.

In the GDB project, we write what I believe are excellent git commit
messages that go in depth describing the observed problem and the
solution provided by the patch.  That provides valuable information for
anyone wanting to understand why the code is the way it is.  A list of
modified symbols doesn't help.  So I am completely fine if the "list of
modified symbols" part of the information is lost.

> If you use Emacs to make the changes
> themselves and commit the changes into Git, you just need to use the
> Emacs commands to format the Git log messages, those commands will
> produce the same information as the manual ChangeLog files
> automatically.

I don't really understand how that helps, you still have to spell out
the symbols.  But please don't assume everyone uses Emacs.  Making a
decision based on "Emacs makes it easy" and ignoring the rest of users
wouldn't be fair.

> If that is not an option that this community wants to adopt, then
> gitlog-to-changelog is probably not the adequate solution, and we
> should try using vcs-to-changelog.py (also in Gnulib).

I tried vcs-to-changelog, it gave horrible/useless results with our
codebase.  This is not an option.

Simon

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Cc: gdb@sourceware.org
> From: Simon Marchi <simon.marchi@polymtl.ca>
> Date: Wed, 17 Mar 2021 11:52:49 -0400
> 
> In the GDB project, we write what I believe are excellent git commit
> messages that go in depth describing the observed problem and the
> solution provided by the patch.  That provides valuable information for
> anyone wanting to understand why the code is the way it is.  A list of
> modified symbols doesn't help.  So I am completely fine if the "list of
> modified symbols" part of the information is lost.

Well, if that's what the majority here wants, then so be it.

(The importance of having the list of modified symbols in the log is
that then one doesn't need advanced Git commands to find out which
changes modified a given function and why.)

> > If that is not an option that this community wants to adopt, then
> > gitlog-to-changelog is probably not the adequate solution, and we
> > should try using vcs-to-changelog.py (also in Gnulib).
> 
> I tried vcs-to-changelog, it gave horrible/useless results with our
> codebase.  This is not an option.

Too bad.  Maybe we should report this to the developer of the script,
it could help fix those shortcomings in the future.

Re: sim: replacing ChangeLog files with online git logs

[Luis Machado]

On 3/17/21 1:09 PM, Eli Zaretskii wrote:
>> Cc: gdb@sourceware.org
>> From: Simon Marchi <simon.marchi@polymtl.ca>
>> Date: Wed, 17 Mar 2021 11:52:49 -0400
>>
>> In the GDB project, we write what I believe are excellent git commit
>> messages that go in depth describing the observed problem and the
>> solution provided by the patch.  That provides valuable information for
>> anyone wanting to understand why the code is the way it is.  A list of
>> modified symbols doesn't help.  So I am completely fine if the "list of
>> modified symbols" part of the information is lost.
> 
> Well, if that's what the majority here wants, then so be it.

That's what we should assess. From chatting with other active GDB 
developers, my feeling is that most of us want to drop the process of 
having to write ChangeLog entries manually. But we tend to keep quiet 
and carry on doing it.

> 
> (The importance of having the list of modified symbols in the log is
> that then one doesn't need advanced Git commands to find out which
> changes modified a given function and why.)
> 

I understand the concern about git. I used to find git a bit too cryptic 
too, but using it daily has made that better.

Now git log/git blame shows very useful information when I'm looking for 
specific changes from a commit, and I rarely need to go through 
ChangeLogs other than to find commits that touched a particular 
function/variable.

The auto-generated ChangeLog entry may still help with that, but it 
would be auto-generated instead of taking time away from active developers.

>>> If that is not an option that this community wants to adopt, then
>>> gitlog-to-changelog is probably not the adequate solution, and we
>>> should try using vcs-to-changelog.py (also in Gnulib).
>>
>> I tried vcs-to-changelog, it gave horrible/useless results with our
>> codebase.  This is not an option.
> 
> Too bad.  Maybe we should report this to the developer of the script,
> it could help fix those shortcomings in the future.
> 

Maybe. But looking into the future, parsing C++ to extract that kind of 
information is really not trivial. So it may never work in a reasonable 
way for GDB without some serious effort put into the script.

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Cc: gdb@sourceware.org, Mike Frysinger <vapier@gentoo.org>
> From: Luis Machado <luis.machado@linaro.org>
> Date: Wed, 17 Mar 2021 14:24:27 -0300
> 
> > Well, if that's what the majority here wants, then so be it.
> 
> That's what we should assess. From chatting with other active GDB 
> developers, my feeling is that most of us want to drop the process of 
> having to write ChangeLog entries manually. But we tend to keep quiet 
> and carry on doing it.

I didn't start this discussion, mind you.

> > (The importance of having the list of modified symbols in the log is
> > that then one doesn't need advanced Git commands to find out which
> > changes modified a given function and why.)
> > 
> 
> I understand the concern about git. I used to find git a bit too cryptic 
> too, but using it daily has made that better.
> 
> Now git log/git blame shows very useful information when I'm looking for 
> specific changes from a commit, and I rarely need to go through 
> ChangeLogs other than to find commits that touched a particular 
> function/variable.

IME "git log" and "git blame" have shortcomings when used for
forensics, and having a ChangeLog-style list of changes helps overcome
that in many important use cases.

> >> I tried vcs-to-changelog, it gave horrible/useless results with our
> >> codebase.  This is not an option.
> > 
> > Too bad.  Maybe we should report this to the developer of the script,
> > it could help fix those shortcomings in the future.
> > 
> 
> Maybe. But looking into the future, parsing C++ to extract that kind of 
> information is really not trivial. So it may never work in a reasonable 
> way for GDB without some serious effort put into the script.

We don't need it to do a perfect job, only a reasonable one.  A 80%
success is a very big step forward wrt not having the information at
all.

Re: sim: replacing ChangeLog files with online git logs

[Luis Machado]

On 3/17/21 2:53 PM, Eli Zaretskii wrote:
>> Cc: gdb@sourceware.org, Mike Frysinger <vapier@gentoo.org>
>> From: Luis Machado <luis.machado@linaro.org>
>> Date: Wed, 17 Mar 2021 14:24:27 -0300
>>
>>> Well, if that's what the majority here wants, then so be it.
>>
>> That's what we should assess. From chatting with other active GDB
>> developers, my feeling is that most of us want to drop the process of
>> having to write ChangeLog entries manually. But we tend to keep quiet
>> and carry on doing it.
> 
> I didn't start this discussion, mind you.
> 

I know. But this has been bothering contributors for a few years now.

I don't want to hijack the sim/ChangeLog thread, but it feels important 
to improve this situation in some way. Given sim/ is under GDB's 
umbrella, might as well consider the whole project in my opinion.

>>> (The importance of having the list of modified symbols in the log is
>>> that then one doesn't need advanced Git commands to find out which
>>> changes modified a given function and why.)
>>>
>>
>> I understand the concern about git. I used to find git a bit too cryptic
>> too, but using it daily has made that better.
>>
>> Now git log/git blame shows very useful information when I'm looking for
>> specific changes from a commit, and I rarely need to go through
>> ChangeLogs other than to find commits that touched a particular
>> function/variable.
> 
> IME "git log" and "git blame" have shortcomings when used for
> forensics, and having a ChangeLog-style list of changes helps overcome
> that in many important use cases.
> 

Right. That's true. Both have their pros and cons. But the git commands, 
IMO, demand no extra time from developers. ChangeLog's, on the other 
hand, do.

A contributor's time spent on the project is a limited and valuable 
resource. Making the best use of that resource is important.

>>>> I tried vcs-to-changelog, it gave horrible/useless results with our
>>>> codebase.  This is not an option.
>>>
>>> Too bad.  Maybe we should report this to the developer of the script,
>>> it could help fix those shortcomings in the future.
>>>
>>
>> Maybe. But looking into the future, parsing C++ to extract that kind of
>> information is really not trivial. So it may never work in a reasonable
>> way for GDB without some serious effort put into the script.
> 
> We don't need it to do a perfect job, only a reasonable one.  A 80%
> success is a very big step forward wrt not having the information at
> all.
> 

Sounds reasonable. But for contributors not seeing value in ChangeLog 
entries, it is hard to justify putting effort/time into such a script.

But I agree it would be nice to have improvements to a script/tool like 
that. It might make sense for someone who sees value in such data to 
invest some time improving its generation.

Re: sim: replacing ChangeLog files with online git logs

[Andrew Burgess]

* Eli Zaretskii via Gdb <gdb@sourceware.org> [2021-03-17 19:53:09 +0200]:

> > Cc: gdb@sourceware.org, Mike Frysinger <vapier@gentoo.org>
> > From: Luis Machado <luis.machado@linaro.org>
> > Date: Wed, 17 Mar 2021 14:24:27 -0300
> > 
> > > Well, if that's what the majority here wants, then so be it.
> > 
> > That's what we should assess. From chatting with other active GDB 
> > developers, my feeling is that most of us want to drop the process of 
> > having to write ChangeLog entries manually. But we tend to keep quiet 
> > and carry on doing it.
> 
> I didn't start this discussion, mind you.
> 
> > > (The importance of having the list of modified symbols in the log is
> > > that then one doesn't need advanced Git commands to find out which
> > > changes modified a given function and why.)
> > > 
> > 
> > I understand the concern about git. I used to find git a bit too cryptic 
> > too, but using it daily has made that better.
> > 
> > Now git log/git blame shows very useful information when I'm looking for 
> > specific changes from a commit, and I rarely need to go through 
> > ChangeLogs other than to find commits that touched a particular 
> > function/variable.
> 
> IME "git log" and "git blame" have shortcomings when used for
> forensics, and having a ChangeLog-style list of changes helps overcome
> that in many important use cases.

Hi Eli,

I'd love to hear more about these many important use cases as this
might help me improve my workflow.  I can't recall ever referring to
the ChangeLogs to help track down a bug, so I'm worried I might be
missing out on some excellent techniques here.

I'm always keen to learn!

Thanks,
Andrew

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Date: Thu, 18 Mar 2021 09:31:47 +0000
> From: Andrew Burgess <andrew.burgess@embecosm.com>
> Cc: Luis Machado <luis.machado@linaro.org>, gdb@sourceware.org
> 
> > IME "git log" and "git blame" have shortcomings when used for
> > forensics, and having a ChangeLog-style list of changes helps overcome
> > that in many important use cases.
> 
> Hi Eli,
> 
> I'd love to hear more about these many important use cases as this
> might help me improve my workflow.  I can't recall ever referring to
> the ChangeLogs to help track down a bug, so I'm worried I might be
> missing out on some excellent techniques here.
> 
> I'm always keen to learn!

I'm okay with discussing this, but the comment up-thread to the effect
that I'd better let the sleeping dogs lie turned me off to some
extent.  I don't want to appear as someone who forces an unwanted
discussion on the community.

But if this is a welcome subject, I'd be happy to share my experience,
FWIW.

Re: sim: replacing ChangeLog files with online git logs

[Maciej W. Rozycki]

On Thu, 18 Mar 2021, Eli Zaretskii via Gdb wrote:

> > I'd love to hear more about these many important use cases as this
> > might help me improve my workflow.  I can't recall ever referring to
> > the ChangeLogs to help track down a bug, so I'm worried I might be
> > missing out on some excellent techniques here.
> > 
> > I'm always keen to learn!
> 
> I'm okay with discussing this, but the comment up-thread to the effect
> that I'd better let the sleeping dogs lie turned me off to some
> extent.  I don't want to appear as someone who forces an unwanted
> discussion on the community.
> 
> But if this is a welcome subject, I'd be happy to share my experience,
> FWIW.

 For the record the glibc project has a way to automatically produce 
ChangeLog from commits at release time.  This may be something to look 
into if useful.

  Maciej

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Date: Mon, 22 Mar 2021 03:19:10 +0100 (CET)
> From: "Maciej W. Rozycki" <macro@orcam.me.uk>
> cc: Andrew Burgess <andrew.burgess@embecosm.com>, gdb@sourceware.org
> 
>  For the record the glibc project has a way to automatically produce 
> ChangeLog from commits at release time.  This may be something to look 
> into if useful.

That's the other script in Gnulib that was mentioned in this
discussion.  Someone said it cannot support our codebase well enough.

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 22 Mar 2021 05:38, Eli Zaretskii via Gdb wrote:
> > Date: Mon, 22 Mar 2021 03:19:10 +0100 (CET)
> > From: "Maciej W. Rozycki" <macro@orcam.me.uk>
> > cc: Andrew Burgess <andrew.burgess@embecosm.com>, gdb@sourceware.org
> > 
> >  For the record the glibc project has a way to automatically produce 
> > ChangeLog from commits at release time.  This may be something to look 
> > into if useful.
> 
> That's the other script in Gnulib that was mentioned in this
> discussion.  Someone said it cannot support our codebase well enough.

vcs-to-changelog assumes it's generating a ChangeLog for the entire git
repo, not a subdir.  that's why it works for glibc, and really doesn't
work for us.  unless we got agreement across all projects in here, and
i didn't want to open that can of worms.

i glanced through the script before to see if it could be adjusted to
limit its path, but it didn't seem easy to pull off, so i gave up.

not that i've evaluated it beyond that (i.e. what format it generates).
-mike

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 17 Mar 2021 15:14, Eli Zaretskii wrote:
> From: Mike Frysinger <vapier@gentoo.org>
> > > That's a different conclusions.  You asked what people are doing with
> > > the ChangeLogs, not whether it was mandatory to keep them.
> > 
> > that's fair.  my point/goal was that if there aren't requirements for it,
> > and there isn't a compelling use case for keeping them, then the easiest
> > thing is to do nothing and delete them.
> > 
> > it seems that there isn't a requirement ->
> 
> Indeed, there's no mandatory requirement to keep the files, not
> anymore.
> 
> > i'm skeptical that the logs add value as anyone who is going to contribute
> > by diving into the source history would be online w/git access ->
> > we punt the logs and stop wasting lots of developer time.
> 
> I'm not sure I agree with your skepticism.  And OTOH, producing the
> ChangeLog from Git logs takes mere seconds and wastes no development
> time.  So I think on balance there's very little price to pay for
> something that some people may value.  Why refuse to pay that small
> price and humor those few people?

Andrew: you OK with me merging my patches ?
https://sourceware.org/pipermail/gdb-patches/2021-January/174894.html
https://sourceware.org/pipermail/gdb-patches/2021-January/174895.html

i feel like we can & should make progress here: generating the files
for release allows us to stop hand writing these entirely.  we can
then spend time debating the format (or whether to generate them) and
not spend any time on hand writing.
-mike

Re: sim: replacing ChangeLog files with online git logs

[Andrew Burgess]

* Mike Frysinger <vapier@gentoo.org> [2021-03-23 01:05:52 -0400]:

> On 17 Mar 2021 15:14, Eli Zaretskii wrote:
> > From: Mike Frysinger <vapier@gentoo.org>
> > > > That's a different conclusions.  You asked what people are doing with
> > > > the ChangeLogs, not whether it was mandatory to keep them.
> > > 
> > > that's fair.  my point/goal was that if there aren't requirements for it,
> > > and there isn't a compelling use case for keeping them, then the easiest
> > > thing is to do nothing and delete them.
> > > 
> > > it seems that there isn't a requirement ->
> > 
> > Indeed, there's no mandatory requirement to keep the files, not
> > anymore.
> > 
> > > i'm skeptical that the logs add value as anyone who is going to contribute
> > > by diving into the source history would be online w/git access ->
> > > we punt the logs and stop wasting lots of developer time.
> > 
> > I'm not sure I agree with your skepticism.  And OTOH, producing the
> > ChangeLog from Git logs takes mere seconds and wastes no development
> > time.  So I think on balance there's very little price to pay for
> > something that some people may value.  Why refuse to pay that small
> > price and humor those few people?
> 
> Andrew: you OK with me merging my patches ?
> https://sourceware.org/pipermail/gdb-patches/2021-January/174894.html
> https://sourceware.org/pipermail/gdb-patches/2021-January/174895.html
> 
> i feel like we can & should make progress here: generating the files
> for release allows us to stop hand writing these entirely.  we can
> then spend time debating the format (or whether to generate them) and
> not spend any time on hand writing.

Sure, if that's what you want, go for it.  I still don't
see/understand the value of the auto-generated files, but I'm coming
to the conclusion that I might never understand.

Thanks,
Andrew

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 23 Mar 2021 11:06, Andrew Burgess wrote:
> Mike Frysinger <vapier@gentoo.org> [2021-03-23 01:05:52 -0400]:
> > On 17 Mar 2021 15:14, Eli Zaretskii wrote:
> > > From: Mike Frysinger <vapier@gentoo.org>
> > > > > That's a different conclusions.  You asked what people are doing with
> > > > > the ChangeLogs, not whether it was mandatory to keep them.
> > > > 
> > > > that's fair.  my point/goal was that if there aren't requirements for it,
> > > > and there isn't a compelling use case for keeping them, then the easiest
> > > > thing is to do nothing and delete them.
> > > > 
> > > > it seems that there isn't a requirement ->
> > > 
> > > Indeed, there's no mandatory requirement to keep the files, not
> > > anymore.
> > > 
> > > > i'm skeptical that the logs add value as anyone who is going to contribute
> > > > by diving into the source history would be online w/git access ->
> > > > we punt the logs and stop wasting lots of developer time.
> > > 
> > > I'm not sure I agree with your skepticism.  And OTOH, producing the
> > > ChangeLog from Git logs takes mere seconds and wastes no development
> > > time.  So I think on balance there's very little price to pay for
> > > something that some people may value.  Why refuse to pay that small
> > > price and humor those few people?
> > 
> > Andrew: you OK with me merging my patches ?
> > https://sourceware.org/pipermail/gdb-patches/2021-January/174894.html
> > https://sourceware.org/pipermail/gdb-patches/2021-January/174895.html
> > 
> > i feel like we can & should make progress here: generating the files
> > for release allows us to stop hand writing these entirely.  we can
> > then spend time debating the format (or whether to generate them) and
> > not spend any time on hand writing.
> 
> Sure, if that's what you want, go for it.  I still don't
> see/understand the value of the auto-generated files, but I'm coming
> to the conclusion that I might never understand.

i don't disagree, but i'm trying to make progress without getting stuck
in the same rut we've been for years ;).

Eli: those patches look OK to you too ?
-mike

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Date: Tue, 23 Mar 2021 23:45:24 -0400
> From: Mike Frysinger <vapier@gentoo.org>
> Cc: gdb@sourceware.org, Andrew Burgess <andrew.burgess@embecosm.com>
> 
> Eli: those patches look OK to you too ?

They just add the script to our repository, AFAICT, so I of course
have no objections.

Re: sim: replacing ChangeLog files with online git logs

[Andrew Burgess]

* Eli Zaretskii <eliz@gnu.org> [2021-03-24 18:46:38 +0200]:

> > Date: Tue, 23 Mar 2021 23:45:24 -0400
> > From: Mike Frysinger <vapier@gentoo.org>
> > Cc: gdb@sourceware.org, Andrew Burgess <andrew.burgess@embecosm.com>
> > 
> > Eli: those patches look OK to you too ?
> 
> They just add the script to our repository, AFAICT, so I of course
> have no objections.

No the second proposes to replace the current ChangeLog with one
auto-generated upon release.

This ChangeLog would in effect be a reformatted `git log` output.

For the simulator this would mean there would no longer be any file
that contained output in the format as:

	* file.c (symbol): description.

As Mike already does not include his ChangeLogs in the commit message
(as we do for the rest of gdb/, etc), and the motivation for this
change is to stop writing any text in that format, so likely future
sim/ contributions would not include text in that format either.

Thanks,
Andrew

Re: sim: replacing ChangeLog files with online git logs

[Eli Zaretskii]

> Date: Wed, 24 Mar 2021 16:59:18 +0000
> From: Andrew Burgess <andrew.burgess@embecosm.com>
> Cc: Mike Frysinger <vapier@gentoo.org>, gdb@sourceware.org
> 
> * Eli Zaretskii <eliz@gnu.org> [2021-03-24 18:46:38 +0200]:
> 
> > > Date: Tue, 23 Mar 2021 23:45:24 -0400
> > > From: Mike Frysinger <vapier@gentoo.org>
> > > Cc: gdb@sourceware.org, Andrew Burgess <andrew.burgess@embecosm.com>
> > > 
> > > Eli: those patches look OK to you too ?
> > 
> > They just add the script to our repository, AFAICT, so I of course
> > have no objections.
> 
> No the second proposes to replace the current ChangeLog with one
> auto-generated upon release.
> 
> This ChangeLog would in effect be a reformatted `git log` output.

I don't object to that, either.  The question is: what we will require
the Git log messages to contain then?  Currently, the entire ChangeLog
entry goes there, but once the manually written ChangeLog files are
retired, what will remain?  That was the crux of the recent
discussion.

> For the simulator this would mean there would no longer be any file
> that contained output in the format as:
> 
> 	* file.c (symbol): description.

Well, I personally will lament the loss of that information.  It's up
toy you-all to decide how much does that weigh.

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 24 Mar 2021 19:39, Eli Zaretskii wrote:
> From: Andrew Burgess <andrew.burgess@embecosm.com>
> > Eli Zaretskii <eliz@gnu.org> [2021-03-24 18:46:38 +0200]:
> > > From: Mike Frysinger <vapier@gentoo.org>
> > > > Eli: those patches look OK to you too ?
> > > 
> > > They just add the script to our repository, AFAICT, so I of course
> > > have no objections.
> > 
> > No the second proposes to replace the current ChangeLog with one
> > auto-generated upon release.
> > 
> > This ChangeLog would in effect be a reformatted `git log` output.
> 
> I don't object to that, either.  The question is: what we will require
> the Git log messages to contain then?  Currently, the entire ChangeLog
> entry goes there, but once the manually written ChangeLog files are
> retired, what will remain?  That was the crux of the recent
> discussion.

we won't require the ChangeLog-like entries in the commit message for sim/.
the generated ChangeLog will have name+date+git log.
-mike

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 24 Mar 2021 14:44, Mike Frysinger via Gdb wrote:
> On 24 Mar 2021 19:39, Eli Zaretskii wrote:
> > From: Andrew Burgess <andrew.burgess@embecosm.com>
> > > Eli Zaretskii <eliz@gnu.org> [2021-03-24 18:46:38 +0200]:
> > > > From: Mike Frysinger <vapier@gentoo.org>
> > > > > Eli: those patches look OK to you too ?
> > > > 
> > > > They just add the script to our repository, AFAICT, so I of course
> > > > have no objections.
> > > 
> > > No the second proposes to replace the current ChangeLog with one
> > > auto-generated upon release.
> > > 
> > > This ChangeLog would in effect be a reformatted `git log` output.
> > 
> > I don't object to that, either.  The question is: what we will require
> > the Git log messages to contain then?  Currently, the entire ChangeLog
> > entry goes there, but once the manually written ChangeLog files are
> > retired, what will remain?  That was the crux of the recent
> > discussion.
> 
> we won't require the ChangeLog-like entries in the commit message for sim/.
> the generated ChangeLog will have name+date+git log.

i wanted to give time for people to respond & post any thoughts that they
might have lingering before i moved forward.  since things have gone quiet
now, i guess we've reached the end?  so i'll give another few days before
i land the change for sim/.
-mike

Re: sim: replacing ChangeLog files with online git logs

[Joel Brobecker]

Hi Mike,

> > > > No the second proposes to replace the current ChangeLog with one
> > > > auto-generated upon release.
> > > > 
> > > > This ChangeLog would in effect be a reformatted `git log` output.
> > > 
> > > I don't object to that, either.  The question is: what we will require
> > > the Git log messages to contain then?  Currently, the entire ChangeLog
> > > entry goes there, but once the manually written ChangeLog files are
> > > retired, what will remain?  That was the crux of the recent
> > > discussion.
> > 
> > we won't require the ChangeLog-like entries in the commit message for sim/.
> > the generated ChangeLog will have name+date+git log.
> 
> i wanted to give time for people to respond & post any thoughts that they
> might have lingering before i moved forward.  since things have gone quiet
> now, i guess we've reached the end?  so i'll give another few days before
> i land the change for sim/.

Can you hold off on this for now? We're discussing it within the group
of GDB Global Maintainers, and we need more time to converge.

-- 
Joel

Re: sim: replacing ChangeLog files with online git logs

[Luis Machado]

Hi Joel,

On 3/31/21 12:27 AM, Joel Brobecker wrote:
> Hi Mike,
> 
>>>>> No the second proposes to replace the current ChangeLog with one
>>>>> auto-generated upon release.
>>>>>
>>>>> This ChangeLog would in effect be a reformatted `git log` output.
>>>>
>>>> I don't object to that, either.  The question is: what we will require
>>>> the Git log messages to contain then?  Currently, the entire ChangeLog
>>>> entry goes there, but once the manually written ChangeLog files are
>>>> retired, what will remain?  That was the crux of the recent
>>>> discussion.
>>>
>>> we won't require the ChangeLog-like entries in the commit message for sim/.
>>> the generated ChangeLog will have name+date+git log.
>>
>> i wanted to give time for people to respond & post any thoughts that they
>> might have lingering before i moved forward.  since things have gone quiet
>> now, i guess we've reached the end?  so i'll give another few days before
>> i land the change for sim/.
> 
> Can you hold off on this for now? We're discussing it within the group
> of GDB Global Maintainers, and we need more time to converge.
> 

Out of curiosity, is the discussion specifically focused on sim/ or 
should we address the bigger issue of ChangeLog's within the GDB codebase?

Re: sim: replacing ChangeLog files with online git logs

[Joel Brobecker]

Hi Luis,

> > Can you hold off on this for now? We're discussing it within the group
> > of GDB Global Maintainers, and we need more time to converge.
> > 
> 
> Out of curiosity, is the discussion specifically focused on sim/ or
> should we address the bigger issue of ChangeLog's within the GDB
> codebase?

The latter (all the GDB codebase).

-- 
Joel

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 31 Mar 2021 07:27, Joel Brobecker wrote:
> > > > > No the second proposes to replace the current ChangeLog with one
> > > > > auto-generated upon release.
> > > > > 
> > > > > This ChangeLog would in effect be a reformatted `git log` output.
> > > > 
> > > > I don't object to that, either.  The question is: what we will require
> > > > the Git log messages to contain then?  Currently, the entire ChangeLog
> > > > entry goes there, but once the manually written ChangeLog files are
> > > > retired, what will remain?  That was the crux of the recent
> > > > discussion.
> > > 
> > > we won't require the ChangeLog-like entries in the commit message for sim/.
> > > the generated ChangeLog will have name+date+git log.
> > 
> > i wanted to give time for people to respond & post any thoughts that they
> > might have lingering before i moved forward.  since things have gone quiet
> > now, i guess we've reached the end?  so i'll give another few days before
> > i land the change for sim/.
> 
> Can you hold off on this for now? We're discussing it within the group
> of GDB Global Maintainers, and we need more time to converge.

certainly.  i just didn't want this to go dead for another 2 years before
restarting the whole thing from scratch.  manually writing ChangeLog entries
anywhere is an incredible waste of my time.
-mike

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 31 Mar 2021 23:41, Mike Frysinger via Gdb wrote:
> On 31 Mar 2021 07:27, Joel Brobecker wrote:
> > > > > > No the second proposes to replace the current ChangeLog with one
> > > > > > auto-generated upon release.
> > > > > > 
> > > > > > This ChangeLog would in effect be a reformatted `git log` output.
> > > > > 
> > > > > I don't object to that, either.  The question is: what we will require
> > > > > the Git log messages to contain then?  Currently, the entire ChangeLog
> > > > > entry goes there, but once the manually written ChangeLog files are
> > > > > retired, what will remain?  That was the crux of the recent
> > > > > discussion.
> > > > 
> > > > we won't require the ChangeLog-like entries in the commit message for sim/.
> > > > the generated ChangeLog will have name+date+git log.
> > > 
> > > i wanted to give time for people to respond & post any thoughts that they
> > > might have lingering before i moved forward.  since things have gone quiet
> > > now, i guess we've reached the end?  so i'll give another few days before
> > > i land the change for sim/.
> > 
> > Can you hold off on this for now? We're discussing it within the group
> > of GDB Global Maintainers, and we need more time to converge.
> 
> certainly.  i just didn't want this to go dead for another 2 years before
> restarting the whole thing from scratch.  manually writing ChangeLog entries
> anywhere is an incredible waste of my time.

how goes it ?
-mike

Re: sim: replacing ChangeLog files with online git logs

[Joel Brobecker]

Hi Mike,

> > > Can you hold off on this for now? We're discussing it within the group
> > > of GDB Global Maintainers, and we need more time to converge.
> how goes it ?

We're making progress. But is often the case with a group of volunteers
working via email, things take time. It's an important decision, so
it's important we take the time needed to finalize this.

-- 
Joel

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 19 Apr 2021 06:32, Joel Brobecker wrote:
> > > > Can you hold off on this for now? We're discussing it within the group
> > > > of GDB Global Maintainers, and we need more time to converge.
> > how goes it ?
> 
> We're making progress. But is often the case with a group of volunteers
> working via email, things take time. It's an important decision, so
> it's important we take the time needed to finalize this.

i get that.  i'll note that this has been ongoing for at least 6.5 years now.
i really wish i could get back all that time spent writing these things.

if you want me to check back in a month, i can.
-mike

Re: sim: replacing ChangeLog files with online git logs

[Joel Brobecker]

> > > > > Can you hold off on this for now? We're discussing it within the group
> > > > > of GDB Global Maintainers, and we need more time to converge.
> > > how goes it ?
> > 
> > We're making progress. But is often the case with a group of volunteers
> > working via email, things take time. It's an important decision, so
> > it's important we take the time needed to finalize this.
> 
> i get that.  i'll note that this has been ongoing for at least 6.5
> years now.  i really wish i could get back all that time spent writing
> these things.

Me too, believe me. The key differences here is that, first, mentalities
have changed; but also, discussions in the past no one took charge of
making the discussion converge, whereas the GDB Global Maintainers
now have. It's just a question of time.

> if you want me to check back in a month, i can.

Please do indeed. I'm hoping that we'll be able to reach a decision
earlier than that, but there's unfortunately limited control on
the timeline.

-- 
Joel

Re: sim: replacing ChangeLog files with online git logs

[Mike Frysinger]

On 19 Apr 2021 09:00, Joel Brobecker wrote:
> > > > > > Can you hold off on this for now? We're discussing it within the group
> > > > > > of GDB Global Maintainers, and we need more time to converge.
> > > > how goes it ?
> > > 
> > > We're making progress. But is often the case with a group of volunteers
> > > working via email, things take time. It's an important decision, so
> > > it's important we take the time needed to finalize this.
> > 
> > i get that.  i'll note that this has been ongoing for at least 6.5
> > years now.  i really wish i could get back all that time spent writing
> > these things.
> 
> Me too, believe me. The key differences here is that, first, mentalities
> have changed; but also, discussions in the past no one took charge of
> making the discussion converge, whereas the GDB Global Maintainers
> now have. It's just a question of time.
> 
> > if you want me to check back in a month, i can.
> 
> Please do indeed. I'm hoping that we'll be able to reach a decision
> earlier than that, but there's unfortunately limited control on
> the timeline.

how's it ?
-mike

Re: sim: replacing ChangeLog files with online git logs

[Luis Machado]

On 5/23/21 12:26 AM, Mike Frysinger via Gdb wrote:
> On 19 Apr 2021 09:00, Joel Brobecker wrote:
>>>>>>> Can you hold off on this for now? We're discussing it within the group
>>>>>>> of GDB Global Maintainers, and we need more time to converge.
>>>>> how goes it ?
>>>>
>>>> We're making progress. But is often the case with a group of volunteers
>>>> working via email, things take time. It's an important decision, so
>>>> it's important we take the time needed to finalize this.
>>>
>>> i get that.  i'll note that this has been ongoing for at least 6.5
>>> years now.  i really wish i could get back all that time spent writing
>>> these things.
>>
>> Me too, believe me. The key differences here is that, first, mentalities
>> have changed; but also, discussions in the past no one took charge of
>> making the discussion converge, whereas the GDB Global Maintainers
>> now have. It's just a question of time.
>>
>>> if you want me to check back in a month, i can.
>>
>> Please do indeed. I'm hoping that we'll be able to reach a decision
>> earlier than that, but there's unfortunately limited control on
>> the timeline.
> 
> how's it ?
> -mike
> 

Ping?

Re: sim: replacing ChangeLog files with online git logs

[Paul Smith]

On Sat, 2021-03-13 at 13:21 -0500, Mike Frysinger via Gdb wrote:
> > > the hard & bikeshed part is what form does the log take.  do you
> > > just want the default `git log` output ?
> > There's a Gnulib script which will produce the form that matches
> > the ChangeLog format.
> 
> there's actually multiple gnulib scripts.  which one are you
> referring to ? gitlog-to-changelog ?

I have no interest in getting in the middle of the discussion on what
is appropriate or not, but FYI this is the text I add to the maintainer
version of the makefile for GNU make, if it helps (the gl2cl-date
variable holds the oldest date to be included in the ChangeLog).


## ---------------------- ##
## Generating ChangeLog.  ##
## ---------------------- ##

gl2cl-date := 2013-10-10
gl2cl := $(GNULIBDIR)/build-aux/gitlog-to-changelog

# Rebuild the changelog whenever a new commit is added
ChangeLog: .check-git-HEAD
	if test -f '$(gl2cl)'; then \
	    '$(gl2cl)' --since='$(gl2cl-date)' > '$@'; \
	else \
	    echo "WARNING: $(gl2cl) is not available.  No $@ generated."; \
	fi

.check-git-HEAD: FORCE
	sha="`git rev-parse HEAD`"; \
	test -f '$@' && [ "`cat '$@' 2>/dev/null`" = "$$sha" ] \
	    || echo "$$sha" > '$@'

.PHONY: FORCE
FORCE:;@: