From martin.densing at psi.ch  Sat Feb 16 13:49:22 2019
From: martin.densing at psi.ch (Densing Martin (PSI))
Date: Sat, 16 Feb 2019 12:49:22 +0000
Subject: [metapost] New numbersystems for the graph package?
Message-ID: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>

Dear metapost experts,

Many thanks to the metapost-coders for the introduction of the new numbersystems!

How can we ensure that the new numbersystems work with the existing, very helpful macros that are available with the usual TeX-disitributions? Most relevant would be the graph macros of John Hobby. Hence: Would it be possible to make the graph macros (and all the auxiliary macros, e.g. format etc.) work with all numbersystems?

Example: 

"mpost -numbersystem=decimal test.mp" gives an error on the following file "test.mp", whereas just "mpost test.mp" runs smoothly. Version: MetaPost 2.000 (TeX Live 2017/W32TeX) on MS Windows.

File: test.mp
% 0  1.4
% 1  3.2
% 2  -4.5
% 3  8.0
% 4  9.0
% 5  4.0

% Data-file for graph can be included at beginning of source-file,
% because %<space>0 1.4 etc. is read by gdata marco until empty line

input graph

prologues := 3;
outputtemplate := "%j-%c.mps";

beginfig(1)  
  draw begingraph(10cm,10cm);
    path p;
    gdata("test.mp", $, augment.p(i, scantokens $2););
    gdraw p;
  endgraph;
endfig;

end
-----------------end file

Metapost's log (if numbersystem=decimal):
-----------------------------------------------------------

! Logarithm of -7000.000523787821841175847573036896 has been replaced by 0.
Gipick_->...i=1upto.mlog(xpart.Gmhh_-xpart.Gmll_)/
....

This is just one of several issues, for example the auxiliary "format" macro has also problems with the new numbersystems.

Thank you very much for your help.

Best,
Martin

__________________________________________
Paul Scherrer Institut 
Martin Densing
Schweiz

E-Mail: martin.densing at psi.ch



From luigi.scarso at gmail.com  Sat Feb 16 16:30:28 2019
From: luigi.scarso at gmail.com (luigi scarso)
Date: Sat, 16 Feb 2019 16:30:28 +0100
Subject: [metapost] New numbersystems for the graph package?
In-Reply-To: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>
References: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>
Message-ID: <CAG5iGsCxsNdwN+Xgp4mQWb5XLnOq1uPx4oL=VRRqYuZPu4VqGA@mail.gmail.com>

On Sat, Feb 16, 2019 at 3:37 PM Densing Martin (PSI) <martin.densing at psi.ch>
wrote:

> Dear metapost experts,
>
> Many thanks to the metapost-coders for the introduction of the new
> numbersystems!
>
> How can we ensure that the new numbersystems work with the existing, very
> helpful macros that are available with the usual TeX-disitributions? Most
> relevant would be the graph macros of John Hobby. Hence: Would it be
> possible to make the graph macros (and all the auxiliary macros, e.g.
> format etc.) work with all numbersystems?
>
> Example:
>
> "mpost -numbersystem=decimal test.mp" gives an error on the following
> file "test.mp", whereas just "mpost test.mp" runs smoothly. Version:
> MetaPost 2.000 (TeX Live 2017/W32TeX) on MS Windows.
>
> File: test.mp
> % 0  1.4
> % 1  3.2
> % 2  -4.5
> % 3  8.0
> % 4  9.0
> % 5  4.0
>
> % Data-file for graph can be included at beginning of source-file,
> % because %<space>0 1.4 etc. is read by gdata marco until empty line
>
> input graph
>
> prologues := 3;
> outputtemplate := "%j-%c.mps";
>
> beginfig(1)
>   draw begingraph(10cm,10cm);
>     path p;
>     gdata("test.mp", $, augment.p(i, scantokens $2););
>     gdraw p;
>   endgraph;
> endfig;
>
> end
> -----------------end file
>
> Metapost's log (if numbersystem=decimal):
> -----------------------------------------------------------
>
> ! Logarithm of -7000.000523787821841175847573036896 has been replaced by 0.
> Gipick_->...i=1upto.mlog(xpart.Gmhh_-xpart.Gmll_)/
> ....
>
> This is just one of several issues, for example the auxiliary "format"
> macro has also problems with the new numbersystems.
>
confirmed , looking into it.

-- 
luigi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/metapost/attachments/20190216/40591f37/attachment.html>

From luigi.scarso at gmail.com  Sat Feb 16 19:48:48 2019
From: luigi.scarso at gmail.com (luigi scarso)
Date: Sat, 16 Feb 2019 19:48:48 +0100
Subject: [metapost] New numbersystems for the graph package?
In-Reply-To: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>
References: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>
Message-ID: <CAG5iGsBGWej0z3aJwy4GFmUD_BU=mwwvg1TBwNG7feH=Bg4xLw@mail.gmail.com>

On Sat, Feb 16, 2019 at 3:37 PM Densing Martin (PSI) <martin.densing at psi.ch>
wrote:

> Dear metapost experts,
>
> Many thanks to the metapost-coders for the introduction of the new
> numbersystems!
>
> How can we ensure that the new numbersystems work with the existing, very
> helpful macros that are available with the usual TeX-disitributions? Most
> relevant would be the graph macros of John Hobby. Hence: Would it be
> possible to make the graph macros (and all the auxiliary macros, e.g.
> format etc.) work with all numbersystems?
>
> Example:
>
> "mpost -numbersystem=decimal test.mp" gives an error on the following
> file "test.mp", whereas just "mpost test.mp" runs smoothly. Version:
> MetaPost 2.000 (TeX Live 2017/W32TeX) on MS Windows.
>
> File: test.mp
> % 0  1.4
> % 1  3.2
> % 2  -4.5
> % 3  8.0
> % 4  9.0
> % 5  4.0
>
> % Data-file for graph can be included at beginning of source-file,
> % because %<space>0 1.4 etc. is read by gdata marco until empty line
>
> input graph
>
> prologues := 3;
> outputtemplate := "%j-%c.mps";
>
> beginfig(1)
>   draw begingraph(10cm,10cm);
>     path p;
>     gdata("test.mp", $, augment.p(i, scantokens $2););
>     gdraw p;
>   endgraph;
> endfig;
>
> end
> -----------------end file
>
> Metapost's log (if numbersystem=decimal):
> -----------------------------------------------------------
>
> ! Logarithm of -7000.000523787821841175847573036896 has been replaced by 0.
> Gipick_->...i=1upto.mlog(xpart.Gmhh_-xpart.Gmll_)/
> ....
>
> This is just one of several issues, for example the auxiliary "format"
> macro has also problems with the new numbersystems.
>
> Thank you very much for your help.
>
>
In
texmf-dist/metapost/base/marith.mp:
vardef Mabs primary x = x*.5*2 enddef;

Can you try with
vardef Mabs primary x = x*1 enddef;


-- 
luigi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/metapost/attachments/20190216/26526552/attachment.html>

From luigi.scarso at gmail.com  Mon Feb 18 01:04:01 2019
From: luigi.scarso at gmail.com (luigi scarso)
Date: Mon, 18 Feb 2019 01:04:01 +0100
Subject: [metapost] New numbersystems for the graph package?
In-Reply-To: <CAG5iGsBGWej0z3aJwy4GFmUD_BU=mwwvg1TBwNG7feH=Bg4xLw@mail.gmail.com>
References: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>
 <CAG5iGsBGWej0z3aJwy4GFmUD_BU=mwwvg1TBwNG7feH=Bg4xLw@mail.gmail.com>
Message-ID: <CAG5iGsDC9oASArRVM9LgQgR9y0ftdJNDJsoz23zaBuB06-DZbA@mail.gmail.com>

On Sat, Feb 16, 2019 at 7:48 PM luigi scarso <luigi.scarso at gmail.com> wrote:

>
>> In
> texmf-dist/metapost/base/marith.mp:
> vardef Mabs primary x = x*.5*2 enddef;
>
> Can you try with
> vardef Mabs primary x = x*1 enddef;
>
> sorry , forget about it .
The correct definition is
vardef Mabs primary x = x*.5*2 enddef;

-- 
luigi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/metapost/attachments/20190218/20fa377c/attachment.html>

From luigi.scarso at gmail.com  Mon Feb 18 01:31:34 2019
From: luigi.scarso at gmail.com (luigi scarso)
Date: Mon, 18 Feb 2019 01:31:34 +0100
Subject: [metapost] New numbersystems for the graph package?
In-Reply-To: <CAG5iGsDC9oASArRVM9LgQgR9y0ftdJNDJsoz23zaBuB06-DZbA@mail.gmail.com>
References: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>
 <CAG5iGsBGWej0z3aJwy4GFmUD_BU=mwwvg1TBwNG7feH=Bg4xLw@mail.gmail.com>
 <CAG5iGsDC9oASArRVM9LgQgR9y0ftdJNDJsoz23zaBuB06-DZbA@mail.gmail.com>
Message-ID: <CAG5iGsAPrN_ZbGMvA-FGP31NAyr1Yc2-TsYPy7nF+xAzCYdhmg@mail.gmail.com>

This should be ok

% 0  1.4
% 1  3.2
% 2  -4.5
% 3  8.0
% 4  9.0
% 5  4.0

% Data-file for graph can be included at beginning of source-file,
% because %<space>0 1.4 etc. is read by gdata marco until empty line

input graph

prologues := 3;
outputtemplate := "%j-%c.mps";
if numbersystem="decimal":
%% numberprecision:=6;
 vardef Mabs primary x = x/10*10 enddef;
fi

beginfig(1)
  draw begingraph(10cm,10cm);
    path p;
    gdata("test.mp", $, augment.p(i, scantokens $2););
    gdraw p;
  endgraph;
endfig;

end

-- 
luigi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/metapost/attachments/20190218/339471e8/attachment.html>

From martin.densing at psi.ch  Mon Feb 18 09:21:04 2019
From: martin.densing at psi.ch (Densing Martin (PSI))
Date: Mon, 18 Feb 2019 08:21:04 +0000
Subject: [metapost] New numbersystems for the graph package?
In-Reply-To: <CAG5iGsBGWej0z3aJwy4GFmUD_BU=mwwvg1TBwNG7feH=Bg4xLw@mail.gmail.com>
References: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>
 <CAG5iGsBGWej0z3aJwy4GFmUD_BU=mwwvg1TBwNG7feH=Bg4xLw@mail.gmail.com>
Message-ID: <7D4340E74A6C8848815563146D838EF8470C7BFB@MBX114.d.ethz.ch>

Dear Luigi,

Thanks for your answer!

This removes the error, but it destroys the y-axis (no negative labels anymore).

>From my limited perspective, the problem seems to be: The graph-macro depends in several places on the finite, original numbersystem because of the Mlog transformation (which would be not needed with higher precision arithmetic). 
Until there is a fix, perhaps an innocent user should be made aware by a conditional errormessage based on the numbersystem at the beginning of the graph-macro, or a note in the documentation?

Best,
Martin


From: luigi scarso <luigi.scarso at gmail.com> 
Sent: Samstag, 16. Februar 2019 19:49
Cc: metapost at tug.org
Subject: Re: [metapost] New numbersystems for the graph package?



On Sat, Feb 16, 2019 at 3:37 PM Densing Martin (PSI) <mailto:martin.densing at psi.ch> wrote:
Dear metapost experts,

Many thanks to the metapost-coders for the introduction of the new numbersystems!

How can we ensure that the new numbersystems work with the existing, very helpful macros that are available with the usual TeX-disitributions? Most relevant would be the graph macros of John Hobby. Hence: Would it be possible to make the graph macros (and all the auxiliary macros, e.g. format etc.) work with all numbersystems?

Example: 

"mpost -numbersystem=decimal http://test.mp" gives an error on the following file "http://test.mp", whereas just "mpost http://test.mp" runs smoothly. Version: MetaPost 2.000 (TeX Live 2017/W32TeX) on MS Windows.

File: http://test.mp
% 0? 1.4
% 1? 3.2
% 2? -4.5
% 3? 8.0
% 4? 9.0
% 5? 4.0

% Data-file for graph can be included at beginning of source-file,
% because %<space>0 1.4 etc. is read by gdata marco until empty line

input graph

prologues := 3;
outputtemplate := "%j-%c.mps";

beginfig(1)? 
? draw begingraph(10cm,10cm);
? ? path p;
? ? gdata("http://test.mp", $, augment.p(i, scantokens $2););
? ? gdraw p;
? endgraph;
endfig;

end
-----------------end file

Metapost's log (if numbersystem=decimal):
-----------------------------------------------------------

! Logarithm of -7000.000523787821841175847573036896 has been replaced by 0.
Gipick_->...i=1upto.mlog(xpart.Gmhh_-xpart.Gmll_)/
....

This is just one of several issues, for example the auxiliary "format" macro has also problems with the new numbersystems.

Thank you very much for your help.
?
In?

texmf-dist/metapost/base/http://marith.mp:?
vardef Mabs primary x = x*.5*2 enddef;

Can you try with?
vardef Mabs primary x = x*1 enddef;


-- 
luigi


From luigi.scarso at gmail.com  Mon Feb 18 09:26:29 2019
From: luigi.scarso at gmail.com (luigi scarso)
Date: Mon, 18 Feb 2019 09:26:29 +0100
Subject: [metapost] New numbersystems for the graph package?
In-Reply-To: <7D4340E74A6C8848815563146D838EF8470C7BFB@MBX114.d.ethz.ch>
References: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>
 <CAG5iGsBGWej0z3aJwy4GFmUD_BU=mwwvg1TBwNG7feH=Bg4xLw@mail.gmail.com>
 <7D4340E74A6C8848815563146D838EF8470C7BFB@MBX114.d.ethz.ch>
Message-ID: <CAG5iGsAJ7LQgwSdvu1ahSSvz3uRvg5Y2swEEqaeFKOOANgbcZw@mail.gmail.com>

On Mon, Feb 18, 2019 at 9:21 AM Densing Martin (PSI) <martin.densing at psi.ch>
wrote:

> Dear Luigi,
>
> Thanks for your answer!
>
> This removes the error, but it destroys the y-axis (no negative labels
> anymore).
>
>
hm if you are using
vardef Mabs primary x = x enddef;
yesm this is wrong, as I have wrote.

But this one should be ok
if numbersystem="decimal":
%% numberprecision:=6;
 vardef Mabs primary x = x/10*10 enddef;
fi



-- 
luigi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/metapost/attachments/20190218/b7611ff7/attachment.html>

From martin.densing at psi.ch  Mon Feb 18 16:51:37 2019
From: martin.densing at psi.ch (Densing Martin (PSI))
Date: Mon, 18 Feb 2019 15:51:37 +0000
Subject: [metapost] New numbersystems for the graph package?
In-Reply-To: <CAG5iGsAPrN_ZbGMvA-FGP31NAyr1Yc2-TsYPy7nF+xAzCYdhmg@mail.gmail.com>
References: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>
 <CAG5iGsBGWej0z3aJwy4GFmUD_BU=mwwvg1TBwNG7feH=Bg4xLw@mail.gmail.com>
 <CAG5iGsDC9oASArRVM9LgQgR9y0ftdJNDJsoz23zaBuB06-DZbA@mail.gmail.com>
 <CAG5iGsAPrN_ZbGMvA-FGP31NAyr1Yc2-TsYPy7nF+xAzCYdhmg@mail.gmail.com>
Message-ID: <7D4340E74A6C8848815563146D838EF8470C7CE3@MBX114.d.ethz.ch>

Dear Luigi,

Thanks you very much. On my computer, the provided test.mp seems still not to work:

The output with numbersystem=decimal is attached (test-1-decimal.mps); the y-axis is still wrong.
The usual output is test-1.mps; the y-axis labelling  is correct (includes negative values).

My point is not this particular example, because I do not need this particular code (by introducing your very good hack).  But still, thank you very much for your good help!
My point is:  It seems that the graph-macros were not tested with the new numbersystems, and because you or Taco Hoekwater have introduced the new numbersystems, it would be nice to have a short comment on this how to proceed regarding the new numbersystems and the graph-macro.

Best regards,
Martin



From: luigi scarso <luigi.scarso at gmail.com>
Sent: Montag, 18. Februar 2019 01:32
Cc: metapost at tug.org
Subject: Re: [metapost] New numbersystems for the graph package?

This should be ok

% 0  1.4
% 1  3.2
% 2  -4.5
% 3  8.0
% 4  9.0
% 5  4.0

% Data-file for graph can be included at beginning of source-file,
% because %<space>0 1.4 etc. is read by gdata marco until empty line

input graph

prologues := 3;
outputtemplate := "%j-%c.mps";
if numbersystem="decimal":
%% numberprecision:=6;
 vardef Mabs primary x = x/10*10 enddef;
fi

beginfig(1)
  draw begingraph(10cm,10cm);
    path p;
    gdata("test.mp<http://test.mp>", $, augment.p(i, scantokens $2););
    gdraw p;
  endgraph;
endfig;

end

--
luigi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/metapost/attachments/20190218/e8839e57/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test-1.mps
Type: application/octet-stream
Size: 35780 bytes
Desc: test-1.mps
URL: <https://tug.org/pipermail/metapost/attachments/20190218/e8839e57/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test-1-decimal.mps
Type: application/octet-stream
Size: 21669 bytes
Desc: test-1-decimal.mps
URL: <https://tug.org/pipermail/metapost/attachments/20190218/e8839e57/attachment-0003.obj>

From luigi.scarso at gmail.com  Mon Feb 18 16:56:33 2019
From: luigi.scarso at gmail.com (luigi scarso)
Date: Mon, 18 Feb 2019 16:56:33 +0100
Subject: [metapost] New numbersystems for the graph package?
In-Reply-To: <7D4340E74A6C8848815563146D838EF8470C7CE3@MBX114.d.ethz.ch>
References: <7D4340E74A6C8848815563146D838EF8470C78ED@MBX114.d.ethz.ch>
 <CAG5iGsBGWej0z3aJwy4GFmUD_BU=mwwvg1TBwNG7feH=Bg4xLw@mail.gmail.com>
 <CAG5iGsDC9oASArRVM9LgQgR9y0ftdJNDJsoz23zaBuB06-DZbA@mail.gmail.com>
 <CAG5iGsAPrN_ZbGMvA-FGP31NAyr1Yc2-TsYPy7nF+xAzCYdhmg@mail.gmail.com>
 <7D4340E74A6C8848815563146D838EF8470C7CE3@MBX114.d.ethz.ch>
Message-ID: <CAG5iGsDu77KyQ_oTw_31_dOySrb_A83p5icNsEuFPnoOsj4eYA@mail.gmail.com>

On Mon, Feb 18, 2019 at 4:52 PM Densing Martin (PSI) <martin.densing at psi.ch>
wrote:

> Dear Luigi,
>
>
>
> Thanks you very much. On my computer, the provided test.mp seems still
> not to work:
>
>
>
> The output with numbersystem=decimal is attached (test-1-decimal.mps); the
> y-axis is still wrong.
>
> The usual output is test-1.mps; the y-axis labelling  is correct (includes
> negative values).
>
>
>
ok, thank you . I will check them later.


> My point is not this particular example, because I do not need this
> particular code (by introducing your very good hack).  But still, thank you
> very much for your good help!
>
> My point is:  It seems that the graph-macros were not tested with the new
> numbersystems, and because you or Taco Hoekwater have introduced the new
> numbersystems, it would be nice to have a short comment on this how to
> proceed regarding the new numbersystems and the graph-macro.
>
>
> Ok I will see what  I can do.
-- 
luigi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/metapost/attachments/20190218/2b0d1325/attachment.html>