[Caml-list] Inconsistent assumptions over interface

[Caml-list] Inconsistent assumptions over interface

[Yann Hamdaoui]

Dear list,

I ran into a problem today while compiling a library of mine. I got the
following error :

"Error: The files [...]/.opam/4.06.0/lib/re/re.cmi and src/format.cmi
make inconsistent assumptions over interface Format".

After a quick search it should be solved by cleaning everything and make
a fresh build. In my case it didn't, and I suspect this is because I
declared a module named Format in my lib which clashes with the
Pervasive.Format that seems to be used by the re library (which is a
dependence). Does it mean that I can't use any XXX as a module name if
it already exists in Pervasives and is used by some library I depend on ?


Y.H


-- 
Caml-list mailing list.  Subscription management and archives:
https://sympa.inria.fr/sympa/arc/caml-list
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
Bug reports: http://caml.inria.fr/bin/caml-bugs

Re: [Caml-list] Inconsistent assumptions over interface

[Nicolás Ojeda Bär]

Hello Yann,

Indeed, you cannot link two different modules of the same name in the
same program, where "same name" essentially means that their .cmi
files are named similarly.
The good news is that the upcoming 4.07 release fixes this
long-standing issue by using long compilation unit names for stdlib
modules which are unlikely to conflict with anything else.
(e.g. the .cmi of the stdlib Format module is named stdlib__format.cmi in 4.07).

Best wishes,
Nicolás

On Sun, Apr 22, 2018 at 4:53 PM, Yann Hamdaoui <yann.hamdaoui@gmail.com> wrote:
> Dear list,
>
> I ran into a problem today while compiling a library of mine. I got the
> following error :
>
> "Error: The files [...]/.opam/4.06.0/lib/re/re.cmi and src/format.cmi
> make inconsistent assumptions over interface Format".
>
> After a quick search it should be solved by cleaning everything and make
> a fresh build. In my case it didn't, and I suspect this is because I
> declared a module named Format in my lib which clashes with the
> Pervasive.Format that seems to be used by the re library (which is a
> dependence). Does it mean that I can't use any XXX as a module name if
> it already exists in Pervasives and is used by some library I depend on ?
>
>
> Y.H
>
>
> --
> Caml-list mailing list.  Subscription management and archives:
> https://sympa.inria.fr/sympa/arc/caml-list
> Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
> Bug reports: http://caml.inria.fr/bin/caml-bugs

-- 
Caml-list mailing list.  Subscription management and archives:
https://sympa.inria.fr/sympa/arc/caml-list
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
Bug reports: http://caml.inria.fr/bin/caml-bugs

Re: [Caml-list] Inconsistent assumptions over interface

[Yann Hamdaoui]

I see. I'm tempted to wait for 4.07 but keeping a module named Format
would still break compatibility for lower versions, so I guess I have to
come up with a different name.
Thank you for your help, Nicolás.

Y.H

On 22/04/2018 17:29, Nicolás Ojeda Bär wrote:
> Hello Yann,
>
> Indeed, you cannot link two different modules of the same name in the
> same program, where "same name" essentially means that their .cmi
> files are named similarly.
> The good news is that the upcoming 4.07 release fixes this
> long-standing issue by using long compilation unit names for stdlib
> modules which are unlikely to conflict with anything else.
> (e.g. the .cmi of the stdlib Format module is named stdlib__format.cmi in 4.07).
>
> Best wishes,
> Nicolás
>
> On Sun, Apr 22, 2018 at 4:53 PM, Yann Hamdaoui <yann.hamdaoui@gmail.com> wrote:
>> Dear list,
>>
>> I ran into a problem today while compiling a library of mine. I got the
>> following error :
>>
>> "Error: The files [...]/.opam/4.06.0/lib/re/re.cmi and src/format.cmi
>> make inconsistent assumptions over interface Format".
>>
>> After a quick search it should be solved by cleaning everything and make
>> a fresh build. In my case it didn't, and I suspect this is because I
>> declared a module named Format in my lib which clashes with the
>> Pervasive.Format that seems to be used by the re library (which is a
>> dependence). Does it mean that I can't use any XXX as a module name if
>> it already exists in Pervasives and is used by some library I depend on ?
>>
>>
>> Y.H
>>
>>
>> --
>> Caml-list mailing list.  Subscription management and archives:
>> https://sympa.inria.fr/sympa/arc/caml-list
>> Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
>> Bug reports: http://caml.inria.fr/bin/caml-bugs



-- 
Caml-list mailing list.  Subscription management and archives:
https://sympa.inria.fr/sympa/arc/caml-list
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
Bug reports: http://caml.inria.fr/bin/caml-bugs

Re: [Caml-list] Inconsistent assumptions over interface

[Yawar Amin]

[-- Attachment #1: Type: text/plain, Size: 2740 bytes --]

Yann, if you use the Dune build tool it automatically prefixes your
package's modules with the package name, e.g. if you have `mypackage/src/
format.ml` in your source, Dune will make sure it gets compiled to
`mypackage__format.cmi`.

Regards,

Yawar

On Mon, Apr 23, 2018 at 10:14 AM, Yann Hamdaoui <yann.hamdaoui@gmail.com>
wrote:

> I see. I'm tempted to wait for 4.07 but keeping a module named Format
> would still break compatibility for lower versions, so I guess I have to
> come up with a different name.
> Thank you for your help, Nicolás.
>
> Y.H
>
> On 22/04/2018 17:29, Nicolás Ojeda Bär wrote:
> > Hello Yann,
> >
> > Indeed, you cannot link two different modules of the same name in the
> > same program, where "same name" essentially means that their .cmi
> > files are named similarly.
> > The good news is that the upcoming 4.07 release fixes this
> > long-standing issue by using long compilation unit names for stdlib
> > modules which are unlikely to conflict with anything else.
> > (e.g. the .cmi of the stdlib Format module is named stdlib__format.cmi
> in 4.07).
> >
> > Best wishes,
> > Nicolás
> >
> > On Sun, Apr 22, 2018 at 4:53 PM, Yann Hamdaoui <yann.hamdaoui@gmail.com>
> wrote:
> >> Dear list,
> >>
> >> I ran into a problem today while compiling a library of mine. I got the
> >> following error :
> >>
> >> "Error: The files [...]/.opam/4.06.0/lib/re/re.cmi and src/format.cmi
> >> make inconsistent assumptions over interface Format".
> >>
> >> After a quick search it should be solved by cleaning everything and make
> >> a fresh build. In my case it didn't, and I suspect this is because I
> >> declared a module named Format in my lib which clashes with the
> >> Pervasive.Format that seems to be used by the re library (which is a
> >> dependence). Does it mean that I can't use any XXX as a module name if
> >> it already exists in Pervasives and is used by some library I depend on
> ?
> >>
> >>
> >> Y.H
> >>
> >>
> >> --
> >> Caml-list mailing list.  Subscription management and archives:
> >> https://sympa.inria.fr/sympa/arc/caml-list
> >> Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
> >> Bug reports: http://caml.inria.fr/bin/caml-bugs
>
>
>
> --
> Caml-list mailing list.  Subscription management and archives:
> https://sympa.inria.fr/sympa/arc/caml-list
> Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
> Bug reports: http://caml.inria.fr/bin/caml-bugs
>

-- 
Caml-list mailing list.  Subscription management and archives:
https://sympa.inria.fr/sympa/arc/caml-list
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
Bug reports: http://caml.inria.fr/bin/caml-bugs

[-- Attachment #2: Type: text/html, Size: 3923 bytes --]

Re: [Caml-list] inconsistent assumptions over interface

[Ashish Agarwal]

[-- Attachment #1: Type: text/plain, Size: 1813 bytes --]

> server code with client code, both seem to have a different
Eliommod_parameters module

Right, thanks! This explains it. There are two Eliommod_parameters
interfaces. We have client/eliom_pervasives.cmi compiled agains the client
one, and _server/foo.cmi compiled against the server one. And then I'm
mixing them somehow.

> Usually in these cases it's a good idea to have a look at the hash of the
inconsistent .cmi files with ocamlobjinfo and try to locate the
corresponding .cmi in the included dirs to get a better idea of what's
going on.

Great suggestion. I'll try this.



On Mon, Apr 27, 2015 at 11:02 AM, Daniel Bünzli <daniel.buenzli@erratique.ch
> wrote:

> > It's Eliom related, though I don't know why that would matter.
> >
> > # Error: The files ~/.opam/4.02.1/lib/eliom/client/eliom_pervasives.cmi
> > # and ../_server/foo.cmi make inconsistent assumptions
> > # over interface Eliommod_parameters
>
> Don't know anything about eliom but it seems that you are trying to
> compile server code with client code, both seem to have a different
> Eliommod_parameters module. I don't know if the Eliommod_parameters cmi are
> supposed to be the same in both cases (if there's a discrepancy that didn't
> use to exist it may be due to compiling with `-keep-locs` which could
> result in problems along the line of [1,2] but I don't know how eliom's
> source is structured).
>
> Usually in these cases it's a good idea to have a look at the hash of the
> inconsistent .cmi files with ocamlobjinfo and try to locate the
> corresponding .cmi in the included dirs to get a better idea of what's
> going on.
>
> Best,
>
> Daniel
>
> [1] https://github.com/ocamllabs/ocaml-ctypes/issues/283
> [2] http://caml.inria.fr/mantis/view.php?id=6311#c13598
>
>
>
>

[-- Attachment #2: Type: text/html, Size: 3129 bytes --]

Re: [Caml-list] inconsistent assumptions over interface

[Daniel Bünzli]

> It's Eliom related, though I don't know why that would matter.
> 
> # Error: The files ~/.opam/4.02.1/lib/eliom/client/eliom_pervasives.cmi
> # and ../_server/foo.cmi make inconsistent assumptions
> # over interface Eliommod_parameters

Don't know anything about eliom but it seems that you are trying to compile server code with client code, both seem to have a different Eliommod_parameters module. I don't know if the Eliommod_parameters cmi are supposed to be the same in both cases (if there's a discrepancy that didn't use to exist it may be due to compiling with `-keep-locs` which could result in problems along the line of [1,2] but I don't know how eliom's source is structured).

Usually in these cases it's a good idea to have a look at the hash of the inconsistent .cmi files with ocamlobjinfo and try to locate the corresponding .cmi in the included dirs to get a better idea of what's going on.

Best,

Daniel

[1] https://github.com/ocamllabs/ocaml-ctypes/issues/283
[2] http://caml.inria.fr/mantis/view.php?id=6311#c13598

Re: [Caml-list] inconsistent assumptions over interface

[Ashish Agarwal]

[-- Attachment #1: Type: text/plain, Size: 1663 bytes --]

Thanks for the suggestion. You're probably right, and so it is Eliom
specific. However, the error message is confusing. The file _server/foo.cmi
is generated. Regardless of what else I'm doing wrong, how could this
single cmi file make inconsistent assumptions with another cmi in the eliom
library.


On Mon, Apr 27, 2015 at 10:53 AM, Drup <drupyog+caml@zoho.com> wrote:

> According to the message, my guess is that you are trying to link a server
> file with a client library while having compiled it with the server library.
> Since client and server libraries don't have the same type, they disagree
> over the interface.
>
> My wild guess would be that your build script is broken.
>
>
> Le 27/04/2015 16:39, Ashish Agarwal a écrit :
>
>> The error "inconsistent assumptions over interface" has always been due
>> to a build artifact not being updated after upgrading some library.
>> Rebuilding one's files from scratch always solves the problem. However, I'm
>> in a situation where I've recompiled everything from scratch many times,
>> but I reproducibly get the error below. Does anyone know what else I should
>> look for?
>>
>> It's Eliom related, though I don't know why that would matter.
>>
>> # Error: The files ~/.opam/4.02.1/lib/eliom/client/eliom_pervasives.cmi
>> #        and ../_server/foo.cmi make inconsistent assumptions
>> #        over interface Eliommod_parameters
>> # *** omake: targets were not rebuilt because of errors:
>> #    _build/lib/_client/foo.cmi
>> #       depends on: _build/lib/foo.eliom
>> #    _build/lib/_client/foo.cmo
>> #       depends on: _build/lib/foo.eliom
>>
>>
>
>

[-- Attachment #2: Type: text/html, Size: 2193 bytes --]

Re: [Caml-list] inconsistent assumptions over interface

[Drup]

According to the message, my guess is that you are trying to link a 
server file with a client library while having compiled it with the 
server library.
Since client and server libraries don't have the same type, they 
disagree over the interface.

My wild guess would be that your build script is broken.

Le 27/04/2015 16:39, Ashish Agarwal a écrit :
> The error "inconsistent assumptions over interface" has always been 
> due to a build artifact not being updated after upgrading some 
> library. Rebuilding one's files from scratch always solves the 
> problem. However, I'm in a situation where I've recompiled everything 
> from scratch many times, but I reproducibly get the error below. Does 
> anyone know what else I should look for?
>
> It's Eliom related, though I don't know why that would matter.
>
> # Error: The files ~/.opam/4.02.1/lib/eliom/client/eliom_pervasives.cmi
> #        and ../_server/foo.cmi make inconsistent assumptions
> #        over interface Eliommod_parameters
> # *** omake: targets were not rebuilt because of errors:
> #    _build/lib/_client/foo.cmi
> #       depends on: _build/lib/foo.eliom
> #    _build/lib/_client/foo.cmo
> #       depends on: _build/lib/foo.eliom
>

[Caml-list] inconsistent assumptions over interface

[Ashish Agarwal]

[-- Attachment #1: Type: text/plain, Size: 821 bytes --]

The error "inconsistent assumptions over interface" has always been due to
a build artifact not being updated after upgrading some library. Rebuilding
one's files from scratch always solves the problem. However, I'm in a
situation where I've recompiled everything from scratch many times, but I
reproducibly get the error below. Does anyone know what else I should look
for?

It's Eliom related, though I don't know why that would matter.

# Error: The files ~/.opam/4.02.1/lib/eliom/client/eliom_pervasives.cmi
#        and ../_server/foo.cmi make inconsistent assumptions
#        over interface Eliommod_parameters
# *** omake: targets were not rebuilt because of errors:
#    _build/lib/_client/foo.cmi
#       depends on: _build/lib/foo.eliom
#    _build/lib/_client/foo.cmo
#       depends on: _build/lib/foo.eliom

[-- Attachment #2: Type: text/html, Size: 1035 bytes --]