gmane.comp.lang.eiffel.ise

RE: Eiffel Vision problem

Emmanuel Stapf — 2012-05-16T04:12:06

Could you create a problem report at http://support.eiffel.com and include the new
wel ECFs as well as the Makefile and Makefile.SH generated in the
EIFGENs/.../W_code directory of the project where the C compilation fails.

Thanks,
Manu



------------------------------------

Re: Eiffel Vision problem

Louis M — 2012-05-16T03:05:49

Hi again,
First of all Thank you Mischa and Manu to take the time to help me. I
really appreciate. The problem is not solve yet. I investigate but I am
still in a dead end. I remove all EIFGENs directory (in precomp and in my
application). I try to precompile directly wel from EiffelStudio (by
opening the Wel project in EiffelStudio and select Precompile). After the
compilation I've got the following error:

ld.exe: cannot find -lshlwapi

I try to found a libshlwapi.a file in my system. I found the file in
$(ISE_EIFFEL)\gcc\windows\mingw\i686-w64-mingw32\. So I try changing the
"-lshlwapi" for
"$(ISE_EIFFEL)\gcc\windows\mingw\i686-w64-mingw32\lib\libshlwapi.a".
Surprisingly, the Vision2 application compile correctly, but when I start
it, a pop-up tell me that the application cannot be open (error 0xc0000005).

That is where I am now. So, if I could again use some of your wisdom, I
dont know what to try next.

Good day,

Louis M


--
View this message in context: http://eiffel.641255.n2.nabble.com/Re-eiffel-s

Re: INTEGER_64 file offsets

Jocelyn Fiat — 2012-05-15T19:12:06

Hi

Years ago, I need to open large file  (huge svn dump)
and I made a some file for that

You can find the code there https://gist.github.com/ab10bd949113b88e13cf

I haven't tried to recompile it lately, so it might not compile out of the box
Just sharing this old code as it is  (draft quality)

INTEGER_64 file offsets

Howard Thomson — 2012-05-15T18:54:31

Hi All,

Is there yet an interface to an open file that has an INTEGER_64 query
for the current file offset, and corresponding seek/go command with an
INTEGER_64 argument ?

Or is the solution to map INTEGER to INTEGER_64 ?

In the case of type mapping, does that apply, necessarily, to the whole
project, or can one type map for specific classes / clusters ?

Regards,

Howard

RE: Eiffel Vision problem

Emmanuel Stapf — 2012-05-15T16:54:26

You need to recompile the code of the precompiled library to take into account the changes Mischa mentioned. To do so, delete the EIFGENs directory in C:\users\louis\documents\eiffel user files\7.0\precomp\spec\windows and recompile your project.

Hope this helps,
Manu



------------------------------------

Re: Eiffel Vision problem

Louis M — 2012-05-15T14:00:35

Thanks Mischa for your answer. But now, I have this error:

gcc.exe: c:\users\louis\documents\eiffel user
files\7.0\precomp\spec\windows\EIFGENs\vision2\W_code/mingw/precomp.lib: No
such file or directory

When I go to the W_code and do à "make" command, I have the following error:

ld.exe: cannot find -lshlwapi
make: *** [precomp.lib] Error 1

So, what can I do with this?

Louis M

On Tue, May 15, 2012 at 12:15 AM, Mischa Megens <mischa.megens-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org>wrote:



--
View this message in context: http://eiffel.641255.n2.nabble.com/Re-eiffel-software-Eiffel-Vision-problem-tp7559964.html
Sent from the Eiffel Software Users mailing list archive at Nabble.com.

[Non-text portions of this message have been removed]



------------------------------------

RE: [eiffel_software] Purpose of once strings

2012-05-15T10:28:03

Hi Carl
 
Once strings were used for quite a while as a way to have string
constants.  Not compile-time constants, but initialized at first use at
run time, possibly with values only known at run time.  While the object
references are constant, their constituent characters are not (they are
as mutable as any other string object) and recently the "readable"
string mechanism has been introduced to fill that void, so to speak.
Another use is to provide a shareable buffer amongst objects of common
lineage (a la zero copy mbufs).  Onces are once in time and space, and
that's a very important and powerful notion.

     R

==================================================
Roger F. Osmond





------------------------------------

Re: Purpose of once strings

Alexander Kogtenkov — 2012-05-15T06:57:05

Hi Carl,


In the early days there were different views on the instructions like

    s := "foo"

The literal "foo" can represent either the same object whenever we
reach it or the creation of a new object. The "once" qualifier removes
any ambiguity.

Apart from the semantic difference these two forms of manifest strings
are slightly different in performance. For example, if there is a loop with
many iterations, one can save on creating new string objects at every
iteration by using "once".

Regards,
Alexander Kogtenkov


------------------------------------

Re: Eiffel Vision problem

Mischa Megens — 2012-05-15T04:15:20

Hi Louis,

I ran into a similar problem in the past, and was advised as follows:

There is an incorrect configuration for mingw in the WEL library. To fix 
this, edit the following two files: $ISE_EIFFEL/library/wel/wel.ecf and 
$ISE_EIFFEL/library/wel/wel-safe.ecf and replace the occurrence of 
"Shlwapi.lib" by "-lshlwapi".

$ISE_EIFFEL represents the installation path of EiffelStudio.

This worked well for me.
Best regards,
Mischa



On 5/14/2012 5:32 PM, Louis M wrote:



[Non-text portions of this message have been removed]



------------------------------------

Eiffel Vision problem

Louis M — 2012-05-15T00:32:03

Hi,
For a long time, I use Eiffel Build on Linux. Everything work fine
there. Today, I try to port one of my applications on Windows. I've
got an error with the precompile library. Here is the problem:

Warning code: VD43

Configuration error: precompiled file is missing or unreadable.
What to do: make sure that the file exists, that appropriate permissions
  are set, and that the precompiled C code has been compiled.

File: c:\users\louis\documents\eiffel user
files\7.0\precomp\spec\windows\EIFGENs\vision2\W_code\mingw\driver.exe

When I check the "C compiler Error", I see this error:

gcc.exe: c:\users\louis\documents\eiffel user
files\7.0\precomp\spec\windows\EIFGENs\vision2\W_code/mingw/precomp.lib:
No such file or directory
gcc.exe: Shlwapi.lib: No such file or directory

So, I try to create I new project with EiffelBuild (the most simple
project EiffelBuild can do). I remove the "EIFGENs" directory in the
precomp directory. After the precompilation, I've got the same
problem. I try this on 3 different

Purpose of once strings

carl94706 — 2012-05-14T16:25:37

I'm curious what you use "once" strings for.

Thanks,
Carl




------------------------------------

RE: Why has READABLE_STRING_8.to_integer but READABLE_STRING_GENERAL hasn't?

Emmanuel Stapf — 2012-05-14T16:19:03

 

I meant in memory storage, i.e. attribute.
 

Currently STRING_32 has no encoding. If you need to encode your strings in a
particular manner, you should use our `encoding' library. The encoding is usually
necessary when you interface with external components.

Manu




------------------------------------

RE: Why has READABLE_STRING_8.to_integer but READABLE_STRING_GENERAL hasn't?

Emmanuel Stapf — 2012-05-14T16:16:20

That's true but it only affects certain type of processing. For example, in a
typical UI application such as EiffelStudio, the impact is quite minimum and
mostly affect our editors which could load large classes.
 

At the moment, we don't have an estimate, it actually depends on a study of what
would be the impact of changing STRING (mutable) to be STRING (immutable) in
existing code.

Manu



------------------------------------

Re: Why has READABLE_STRING_8.to_integer but READABLE_STRING_GENERAL hasn't?

Thomas Beale — 2012-05-13T18:59:30

Manu, what does this statement mean? What 'storage back-end'? We use 
UTF-8 as the native 'storage' of string data, as most systems in non 
character-based languages. Probably I am misunderstanding what you mean.

Can I also double check that STRING 32 only has utility as a UTF-32 
representation - there is no other kind of STRING_32, correct? If 
everything evolves to just UTF-*, with invisible conversions being made 
(presumably via 'convert' section code?) between 8 and 32, that would be 
great. (I suppose UTF-16 also has to be handled, but really it is pretty 
useless. Mixed UTF-8 and UTF-32 makes more sense.)

- thomas



[Non-text portions of this message have been removed]



------------------------------------

Re: Why has READABLE_STRING_8.to_integer but READABLE_STRING_GENERAL hasn't?

Jocelyn Fiat — 2012-05-11T07:07:02

About missing features in *_STRING_GENERAL classes

On the READABLE_STRING_... interface

- _GENERAL has the is_integer_8 and so on , and missing to_integer_8   (and
variant for boolean, and so on) It is always possible to do
as_string_8.to_integer_8 ... but not that nice.

about index_of_code (..), I guess we could also have
- last_index_of_code (...)
- occurrences_of_code (...)

Something that I am missing often
- is_case_insensitive_equal (other: like Current): BOOLEAN


On the writable on, i.e  STRING_GENERAL

- replace_substring(_all)
- left_adjust and right_adjust    (by the way, it would be nice to have
 `left_and_right_adjust' )
- left_justify, center_justify and right_justify
- prepend_boolean (..) and other variants
- append_boolean (..) and other variants
- wipe_out
- mirrored and mirror
- multiply
And using code instead of character_*
- it would be nice to have  prune_code  prune_all_code ... similar to
index_of_code usage
    and prune_all_leading_code, prune_all_trailing_code
- character_justif

Re: Why has READABLE_STRING_8.to_integer but READABLE_STRING_GENERAL hasn't?

Peter Gummer — 2012-05-11T06:05:47


I did not realise (or perhaps I have forgotten) that this was your plan. It's a good plan, because the current arrangement of converting between STRING_8 and STRING_32 is processor-intensive and error-prone. On the other hand, the speed gain may come at the cost of increasing memory usage, perhaps by up to a factor of 4.

When is this improvement likely to happen?

- Peter Gummer

------------------------------------

Re: Why has READABLE_STRING_8.to_integer but READABLE_STRING_GENERAL hasn't?

Berend de Boer — 2012-05-10T21:12:03

    Emmanuel> The recommended way to achieve that is to make sure that
    Emmanuel> all routines taking a string as argument should be
    Emmanuel> changed to READABLE_STRING_GENERAL and the storage
    Emmanuel> backend to a variant of STRING_32.

I frequently have to do READABLE_STRING_GENERAL.out to interface with
old code. It would be beautiful if `out' returned a copy-on-write
STRING.

RE: Why has READABLE_STRING_8.to_integer but READABLE_STRING_GENERAL hasn't?

Emmanuel Stapf — 2012-05-10T20:56:17

This one is not possible as READABLE_STRING_GENERAL knows nothing about
CHARACTER_xx. READABLE_STRING_GENERAL is a sequence of natural `code'. Only the
specialized descendants will use either CHARACTER_8 or CHARACTER_32. So in
READABLE_STRING_GENERAL you can only manipulate natural code. In this particular
case, we provide `index_of_code'.
 

Ultimately we will only have READABLE_STRING, IMMUTABLE_STRING and STRING (i.e.
equivalent to today's READABLE_STRING_32, IMMUTABLE_STRING_32 and STRING_32).
Because we have a lot of code that is still depending on STRING_8 for a proper
behavior we had to come up with solutions that have less impact on our customers
and still be able to upgrade our libraries to Unicode strings.

The recommended way to achieve that is to make sure that all routines taking a
string as argument should be changed to READABLE_STRING_GENERAL and the storage
backend to a variant of STRING_32.

Regards,
Manu




------------------------------------

Re: Why has READABLE_STRING_8.to_integer but READABLE_STRING_GENERAL hasn't?

Berend de Boer — 2012-05-10T20:46:03

    Jocelyn> Would it a be a good idea to review *_STRING_8 and _32,
    Jocelyn> and add as much as possible the features missing from
    Jocelyn> _GENERAL ?

Yeah, had another one: index_of;

I don't really understand READABLE_STRING_GENERAL yet, but I assume
it's a good thing to use right?

And it converts automatically from 8 to 32 bits and back?

Re: Why has READABLE_STRING_8.to_integer but READABLE_STRING_GENERAL hasn't?

Jocelyn Fiat — 2012-05-10T20:25:50

Would it a be a good idea to review *_STRING_8 and _32, and add as much as
possible the features missing from _GENERAL ?

RE: Why has READABLE_STRING_8.to_integer but READABLE_STRING_GENERAL hasn't?

Emmanuel Stapf — 2012-05-10T20:02:24

Yes this can be added.

Manu



------------------------------------