From leo at halfaya.org  Fri Feb  1 18:07:12 2019
From: leo at halfaya.org (John Leo)
Date: Fri, 1 Feb 2019 09:07:12 -0800
Subject: [Agda-dev] COMPILE pragma not allowed in safe mode.
Message-ID: <CAHYM_VnXPHjzkL7+CQbC02-gnFk2KbSb1utKT9XgNRDRsgwjkw@mail.gmail.com>

Hi everyone,

I just pulled the latest agda and agda-stdlib and while agda builds fine I
get a bunch of errors in the stdlib code (for example in Data/Empty.agda):
COMPILE pragma not allowed in safe mode.

I don't know if this is a new check that was added and the stdlib has just
not been updated yet but I thought I'd point it out. If you like I can file
a bug. I've just commented out the offending COMPILE statements in my own
code for now.

John
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.chalmers.se/mailman/private/agda-dev/attachments/20190201/a4af6c38/attachment.html>

From nad at cse.gu.se  Sat Feb  2 21:16:48 2019
From: nad at cse.gu.se (Nils Anders Danielsson)
Date: Sat, 2 Feb 2019 21:16:48 +0100
Subject: [Agda-dev] COMPILE pragma not allowed in safe mode.
In-Reply-To: <CAHYM_VnXPHjzkL7+CQbC02-gnFk2KbSb1utKT9XgNRDRsgwjkw@mail.gmail.com>
References: <CAHYM_VnXPHjzkL7+CQbC02-gnFk2KbSb1utKT9XgNRDRsgwjkw@mail.gmail.com>
Message-ID: <105a3118-5911-050e-522f-cfe81ccf5507@cse.gu.se>

On 01/02/2019 18.07, John Leo wrote:
> I just pulled the latest agda and agda-stdlib and while agda builds
> fine I get a bunch of errors in the stdlib code (for example in
> Data/Empty.agda): COMPILE pragma not allowed in safe mode.
> 
> I don't know if this is a new check that was added and the stdlib has
> just not been updated yet but I thought I'd point it out. If you like
> I can file a bug. I've just commented out the offending COMPILE
> statements in my own code for now.

One part of the Agda test suite is to type-check the standard library.
At the moment this test seems to pass for a standard library commit with
hash 6ed677928b0d7c54521b1d0c7cef7e3181f293a8.

I think the idea is that the experimental branch of the standard library
should be compatible with bleeding edge Agda, but I am not aware of any
automated tests that warn if this property does not hold.

-- 
/NAD

From leo at halfaya.org  Sat Feb  2 21:50:35 2019
From: leo at halfaya.org (John Leo)
Date: Sat, 2 Feb 2019 12:50:35 -0800
Subject: [Agda-dev] COMPILE pragma not allowed in safe mode.
In-Reply-To: <105a3118-5911-050e-522f-cfe81ccf5507@cse.gu.se>
References: <CAHYM_VnXPHjzkL7+CQbC02-gnFk2KbSb1utKT9XgNRDRsgwjkw@mail.gmail.com>
	<105a3118-5911-050e-522f-cfe81ccf5507@cse.gu.se>
Message-ID: <CAHYM_Vm2ikuyuFRZ_VobKupkCda3chwM=mGPjd77RUp7BN47LQ@mail.gmail.com>

Hi Nils,

Thanks for the info. However I don't see that commit on github either in
the master or experimental branch. Perhaps I'm missing something. Could you
send me a web link if you have it?

Also I've been pulling latest agda-stdlib from the master branch to go with
agda from the master branch. Are you saying I should be using experimental
stdlib with master agda?

John

On Sat, Feb 2, 2019 at 12:16 PM Nils Anders Danielsson <nad at cse.gu.se>
wrote:

> On 01/02/2019 18.07, John Leo wrote:
> > I just pulled the latest agda and agda-stdlib and while agda builds
> > fine I get a bunch of errors in the stdlib code (for example in
> > Data/Empty.agda): COMPILE pragma not allowed in safe mode.
> >
> > I don't know if this is a new check that was added and the stdlib has
> > just not been updated yet but I thought I'd point it out. If you like
> > I can file a bug. I've just commented out the offending COMPILE
> > statements in my own code for now.
>
> One part of the Agda test suite is to type-check the standard library.
> At the moment this test seems to pass for a standard library commit with
> hash 6ed677928b0d7c54521b1d0c7cef7e3181f293a8.
>
> I think the idea is that the experimental branch of the standard library
> should be compatible with bleeding edge Agda, but I am not aware of any
> automated tests that warn if this property does not hold.
>
> --
> /NAD
> _______________________________________________
> Agda-dev mailing list
> Agda-dev at lists.chalmers.se
> https://lists.chalmers.se/mailman/listinfo/agda-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.chalmers.se/mailman/private/agda-dev/attachments/20190202/93aa0ce1/attachment.html>

From leo at halfaya.org  Sat Feb  2 22:04:12 2019
From: leo at halfaya.org (John Leo)
Date: Sat, 2 Feb 2019 13:04:12 -0800
Subject: [Agda-dev] COMPILE pragma not allowed in safe mode.
In-Reply-To: <CAHYM_Vm2ikuyuFRZ_VobKupkCda3chwM=mGPjd77RUp7BN47LQ@mail.gmail.com>
References: <CAHYM_VnXPHjzkL7+CQbC02-gnFk2KbSb1utKT9XgNRDRsgwjkw@mail.gmail.com>
	<105a3118-5911-050e-522f-cfe81ccf5507@cse.gu.se>
	<CAHYM_Vm2ikuyuFRZ_VobKupkCda3chwM=mGPjd77RUp7BN47LQ@mail.gmail.com>
Message-ID: <CAHYM_VnV7+RMKXoC5OWck6Urqb7nZZL62NFyQyav-_NFXVFZrw@mail.gmail.com>

Looking further the culprit seems to be this commit:
https://github.com/agda/agda/commit/85f1696c0963ca1c8a1930a97012b0bbcedd4cf6

Meanwhile the latest versions on both master and experimental show
agda-stdlib violating this constraint:
https://github.com/agda/agda-stdlib/blob/master/src/Data/Empty.agda

I don't know what version of stdlib agda is testing against, but this
stdlib change was made 2 months ago.

As I said I can file a bug if people want me to, but I guess this problem
is already known. I just wanted to mention it in case it wasn't.

John

On Sat, Feb 2, 2019 at 12:50 PM John Leo <leo at halfaya.org> wrote:

> Hi Nils,
>
> Thanks for the info. However I don't see that commit on github either in
> the master or experimental branch. Perhaps I'm missing something. Could you
> send me a web link if you have it?
>
> Also I've been pulling latest agda-stdlib from the master branch to go
> with agda from the master branch. Are you saying I should be using
> experimental stdlib with master agda?
>
> John
>
> On Sat, Feb 2, 2019 at 12:16 PM Nils Anders Danielsson <nad at cse.gu.se>
> wrote:
>
>> On 01/02/2019 18.07, John Leo wrote:
>> > I just pulled the latest agda and agda-stdlib and while agda builds
>> > fine I get a bunch of errors in the stdlib code (for example in
>> > Data/Empty.agda): COMPILE pragma not allowed in safe mode.
>> >
>> > I don't know if this is a new check that was added and the stdlib has
>> > just not been updated yet but I thought I'd point it out. If you like
>> > I can file a bug. I've just commented out the offending COMPILE
>> > statements in my own code for now.
>>
>> One part of the Agda test suite is to type-check the standard library.
>> At the moment this test seems to pass for a standard library commit with
>> hash 6ed677928b0d7c54521b1d0c7cef7e3181f293a8.
>>
>> I think the idea is that the experimental branch of the standard library
>> should be compatible with bleeding edge Agda, but I am not aware of any
>> automated tests that warn if this property does not hold.
>>
>> --
>> /NAD
>> _______________________________________________
>> Agda-dev mailing list
>> Agda-dev at lists.chalmers.se
>> https://lists.chalmers.se/mailman/listinfo/agda-dev
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.chalmers.se/mailman/private/agda-dev/attachments/20190202/61731e04/attachment.html>

From guillaume.allais at ens-lyon.org  Sun Feb  3 01:00:01 2019
From: guillaume.allais at ens-lyon.org (Guillaume Allais)
Date: Sun, 3 Feb 2019 01:00:01 +0100
Subject: [Agda-dev] COMPILE pragma not allowed in safe mode.
In-Reply-To: <CAHYM_VnV7+RMKXoC5OWck6Urqb7nZZL62NFyQyav-_NFXVFZrw@mail.gmail.com>
References: <CAHYM_VnXPHjzkL7+CQbC02-gnFk2KbSb1utKT9XgNRDRsgwjkw@mail.gmail.com>
	<105a3118-5911-050e-522f-cfe81ccf5507@cse.gu.se>
	<CAHYM_Vm2ikuyuFRZ_VobKupkCda3chwM=mGPjd77RUp7BN47LQ@mail.gmail.com>
	<CAHYM_VnV7+RMKXoC5OWck6Urqb7nZZL62NFyQyav-_NFXVFZrw@mail.gmail.com>
Message-ID: <d2db7934-9962-d572-429e-5cca9ae80865@ens-lyon.org>

My bad. The CI worked, including the stdlib-based tests (?!) so I assumed
everything was OK. I'll look into updating the experimental branch of the
stdlib.

We're discussing the situation on this re-opened issue:
https://github.com/agda/agda-stdlib/issues/547

Looks like I'll need to add some explicit uses of `--sized-types` too now
that Agda thoroughly checks the OPTIONS a module's dependencies are using.

Cheers,


On 02/02/2019 22:04, John Leo wrote:
> Looking further the culprit seems to be this commit:
> https://github.com/agda/agda/commit/85f1696c0963ca1c8a1930a97012b0bbcedd4cf6
> 
> Meanwhile the latest versions on both master and experimental show
> agda-stdlib violating this constraint:
> https://github.com/agda/agda-stdlib/blob/master/src/Data/Empty.agda
> 
> I don't know what version of stdlib agda is testing against, but this
> stdlib change was made 2 months ago.
> 
> As I said I can file a bug if people want me to, but I guess this problem
> is already known. I just wanted to mention it in case it wasn't.
> 
> John
> 
> On Sat, Feb 2, 2019 at 12:50 PM John Leo <leo at halfaya.org> wrote:
> 
>> Hi Nils,
>>
>> Thanks for the info. However I don't see that commit on github either in
>> the master or experimental branch. Perhaps I'm missing something. Could you
>> send me a web link if you have it?
>>
>> Also I've been pulling latest agda-stdlib from the master branch to go
>> with agda from the master branch. Are you saying I should be using
>> experimental stdlib with master agda?
>>
>> John
>>
>> On Sat, Feb 2, 2019 at 12:16 PM Nils Anders Danielsson <nad at cse.gu.se>
>> wrote:
>>
>>> On 01/02/2019 18.07, John Leo wrote:
>>>> I just pulled the latest agda and agda-stdlib and while agda builds
>>>> fine I get a bunch of errors in the stdlib code (for example in
>>>> Data/Empty.agda): COMPILE pragma not allowed in safe mode.
>>>>
>>>> I don't know if this is a new check that was added and the stdlib has
>>>> just not been updated yet but I thought I'd point it out. If you like
>>>> I can file a bug. I've just commented out the offending COMPILE
>>>> statements in my own code for now.
>>>
>>> One part of the Agda test suite is to type-check the standard library.
>>> At the moment this test seems to pass for a standard library commit with
>>> hash 6ed677928b0d7c54521b1d0c7cef7e3181f293a8.
>>>
>>> I think the idea is that the experimental branch of the standard library
>>> should be compatible with bleeding edge Agda, but I am not aware of any
>>> automated tests that warn if this property does not hold.
>>>
>>> --
>>> /NAD
>>> _______________________________________________
>>> Agda-dev mailing list
>>> Agda-dev at lists.chalmers.se
>>> https://lists.chalmers.se/mailman/listinfo/agda-dev
>>>
>>
> 
> 
> _______________________________________________
> Agda-dev mailing list
> Agda-dev at lists.chalmers.se
> https://lists.chalmers.se/mailman/listinfo/agda-dev
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <https://lists.chalmers.se/mailman/private/agda-dev/attachments/20190203/7d5db072/attachment.sig>

From guillaume.allais at ens-lyon.org  Sun Feb  3 02:04:01 2019
From: guillaume.allais at ens-lyon.org (Guillaume Allais)
Date: Sun, 3 Feb 2019 02:04:01 +0100
Subject: [Agda-dev] COMPILE pragma not allowed in safe mode.
In-Reply-To: <d2db7934-9962-d572-429e-5cca9ae80865@ens-lyon.org>
References: <CAHYM_VnXPHjzkL7+CQbC02-gnFk2KbSb1utKT9XgNRDRsgwjkw@mail.gmail.com>
	<105a3118-5911-050e-522f-cfe81ccf5507@cse.gu.se>
	<CAHYM_Vm2ikuyuFRZ_VobKupkCda3chwM=mGPjd77RUp7BN47LQ@mail.gmail.com>
	<CAHYM_VnV7+RMKXoC5OWck6Urqb7nZZL62NFyQyav-_NFXVFZrw@mail.gmail.com>
	<d2db7934-9962-d572-429e-5cca9ae80865@ens-lyon.org>
Message-ID: <1124b0c9-7732-0507-5f38-1a4bde6b43a3@ens-lyon.org>

I have pushed a fix to the experimental branch. It works on my machine with
agda 2.6.0-f220dda (aka the current HEAD).

On 03/02/2019 01:00, Guillaume Allais wrote:
> My bad. The CI worked, including the stdlib-based tests (?!) so I assumed
> everything was OK. I'll look into updating the experimental branch of the
> stdlib.
> 
> We're discussing the situation on this re-opened issue:
> https://github.com/agda/agda-stdlib/issues/547
> 
> Looks like I'll need to add some explicit uses of `--sized-types` too now
> that Agda thoroughly checks the OPTIONS a module's dependencies are using.
> 
> Cheers,
> 
> 
> On 02/02/2019 22:04, John Leo wrote:
>> Looking further the culprit seems to be this commit:
>> https://github.com/agda/agda/commit/85f1696c0963ca1c8a1930a97012b0bbcedd4cf6
>>
>> Meanwhile the latest versions on both master and experimental show
>> agda-stdlib violating this constraint:
>> https://github.com/agda/agda-stdlib/blob/master/src/Data/Empty.agda
>>
>> I don't know what version of stdlib agda is testing against, but this
>> stdlib change was made 2 months ago.
>>
>> As I said I can file a bug if people want me to, but I guess this problem
>> is already known. I just wanted to mention it in case it wasn't.
>>
>> John
>>
>> On Sat, Feb 2, 2019 at 12:50 PM John Leo <leo at halfaya.org> wrote:
>>
>>> Hi Nils,
>>>
>>> Thanks for the info. However I don't see that commit on github either in
>>> the master or experimental branch. Perhaps I'm missing something. Could you
>>> send me a web link if you have it?
>>>
>>> Also I've been pulling latest agda-stdlib from the master branch to go
>>> with agda from the master branch. Are you saying I should be using
>>> experimental stdlib with master agda?
>>>
>>> John
>>>
>>> On Sat, Feb 2, 2019 at 12:16 PM Nils Anders Danielsson <nad at cse.gu.se>
>>> wrote:
>>>
>>>> On 01/02/2019 18.07, John Leo wrote:
>>>>> I just pulled the latest agda and agda-stdlib and while agda builds
>>>>> fine I get a bunch of errors in the stdlib code (for example in
>>>>> Data/Empty.agda): COMPILE pragma not allowed in safe mode.
>>>>>
>>>>> I don't know if this is a new check that was added and the stdlib has
>>>>> just not been updated yet but I thought I'd point it out. If you like
>>>>> I can file a bug. I've just commented out the offending COMPILE
>>>>> statements in my own code for now.
>>>>
>>>> One part of the Agda test suite is to type-check the standard library.
>>>> At the moment this test seems to pass for a standard library commit with
>>>> hash 6ed677928b0d7c54521b1d0c7cef7e3181f293a8.
>>>>
>>>> I think the idea is that the experimental branch of the standard library
>>>> should be compatible with bleeding edge Agda, but I am not aware of any
>>>> automated tests that warn if this property does not hold.
>>>>
>>>> --
>>>> /NAD
>>>> _______________________________________________
>>>> Agda-dev mailing list
>>>> Agda-dev at lists.chalmers.se
>>>> https://lists.chalmers.se/mailman/listinfo/agda-dev
>>>>
>>>
>>
>>
>> _______________________________________________
>> Agda-dev mailing list
>> Agda-dev at lists.chalmers.se
>> https://lists.chalmers.se/mailman/listinfo/agda-dev
>>
> 
> 
> _______________________________________________
> Agda-dev mailing list
> Agda-dev at lists.chalmers.se
> https://lists.chalmers.se/mailman/listinfo/agda-dev
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <https://lists.chalmers.se/mailman/private/agda-dev/attachments/20190203/c99b8629/attachment.sig>

From nad at cse.gu.se  Mon Feb  4 15:28:53 2019
From: nad at cse.gu.se (Nils Anders Danielsson)
Date: Mon, 4 Feb 2019 15:28:53 +0100
Subject: [Agda-dev] COMPILE pragma not allowed in safe mode.
In-Reply-To: <CAHYM_Vm2ikuyuFRZ_VobKupkCda3chwM=mGPjd77RUp7BN47LQ@mail.gmail.com>
References: <CAHYM_VnXPHjzkL7+CQbC02-gnFk2KbSb1utKT9XgNRDRsgwjkw@mail.gmail.com>
	<105a3118-5911-050e-522f-cfe81ccf5507@cse.gu.se>
	<CAHYM_Vm2ikuyuFRZ_VobKupkCda3chwM=mGPjd77RUp7BN47LQ@mail.gmail.com>
Message-ID: <7fafa0fd-76d7-cb41-80fd-fb543774dd72@cse.gu.se>

On 02/02/2019 21.50, John Leo wrote:
> Also I've been pulling latest agda-stdlib from the master branch to go
> with agda from the master branch. Are you saying I should be using
> experimental stdlib with master agda?

At least that's what the standard library's README says.

-- 
/NAD

From nad at cse.gu.se  Mon Feb  4 15:31:32 2019
From: nad at cse.gu.se (Nils Anders Danielsson)
Date: Mon, 4 Feb 2019 15:31:32 +0100
Subject: [Agda-dev] COMPILE pragma not allowed in safe mode.
In-Reply-To: <d2db7934-9962-d572-429e-5cca9ae80865@ens-lyon.org>
References: <CAHYM_VnXPHjzkL7+CQbC02-gnFk2KbSb1utKT9XgNRDRsgwjkw@mail.gmail.com>
	<105a3118-5911-050e-522f-cfe81ccf5507@cse.gu.se>
	<CAHYM_Vm2ikuyuFRZ_VobKupkCda3chwM=mGPjd77RUp7BN47LQ@mail.gmail.com>
	<CAHYM_VnV7+RMKXoC5OWck6Urqb7nZZL62NFyQyav-_NFXVFZrw@mail.gmail.com>
	<d2db7934-9962-d572-429e-5cca9ae80865@ens-lyon.org>
Message-ID: <85e596c9-9684-0c9e-9468-d195f6fc5606@cse.gu.se>

On 03/02/2019 01.00, Guillaume Allais wrote:
> My bad. The CI worked, including the stdlib-based tests (?!) so I assumed
> everything was OK.

I guess there are two issues to consider:

* Agda's test suite does not automatically track the standard library's
   experimental branch. The standard library submodule is updated
   manually.

* The standard library's master branch is not always merged into the
   experimental branch.

-- 
/NAD

From leo at halfaya.org  Mon Feb  4 15:47:54 2019
From: leo at halfaya.org (John Leo)
Date: Mon, 4 Feb 2019 06:47:54 -0800
Subject: [Agda-dev] COMPILE pragma not allowed in safe mode.
In-Reply-To: <7fafa0fd-76d7-cb41-80fd-fb543774dd72@cse.gu.se>
References: <CAHYM_VnXPHjzkL7+CQbC02-gnFk2KbSb1utKT9XgNRDRsgwjkw@mail.gmail.com>
	<105a3118-5911-050e-522f-cfe81ccf5507@cse.gu.se>
	<CAHYM_Vm2ikuyuFRZ_VobKupkCda3chwM=mGPjd77RUp7BN47LQ@mail.gmail.com>
	<7fafa0fd-76d7-cb41-80fd-fb543774dd72@cse.gu.se>
Message-ID: <CAHYM_VmdZAWV8c1W031MKGBAEgisT_D1cvQ=H4ELBX0H_ddpYw@mail.gmail.com>

Okay, thanks. I'd been using master of both together for a long time and
didn't realize the instructions had changed.

It's a bit confusing, though--it would be better to have the two master
branches try to stay in sync, and similarly have other labeled branches of
the two stay in sync. Experimental sounds to me like something that would
go instead with the adga future branch, or else be for other experimental
independent work.

John

On Mon, Feb 4, 2019 at 6:28 AM Nils Anders Danielsson <nad at cse.gu.se> wrote:

> On 02/02/2019 21.50, John Leo wrote:
> > Also I've been pulling latest agda-stdlib from the master branch to go
> > with agda from the master branch. Are you saying I should be using
> > experimental stdlib with master agda?
>
> At least that's what the standard library's README says.
>
> --
> /NAD
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.chalmers.se/mailman/private/agda-dev/attachments/20190204/7488f1e0/attachment.html>