erlhive-devel — Discussion list for erlhive development

Re: [Erlhive-devel] problem getting started

[Ulf W. <ul...@wi...>]

2007/9/13, Filippo Pacini <pa...@sg...>:
> Hi Ulf,
>
> Ulf Wiger wrote:
> > Hi Filippo,
> >
> > I just tried with the latest version (trunk):
> >
> > Eshell V5.5.4  (abort with ^G)
> It seems a problem with the the latest version of erlang.
>
> I have two versions of erlang installed:
> - v5.5.5 (R11B-5) compiled from sources and here I still get the same error
> - v5.5.2 installed with 'apt-get erlang' works as expected.

I need to install the latest version of Erlang and investigate, then,
but not tonight. (:

BR,
Ulf W

Re: [Erlhive-devel] problem getting started

[Filippo P. <pa...@sg...>]

Hi Ulf,

Ulf Wiger wrote:
> Hi Filippo,
> 
> I just tried with the latest version (trunk):
> 
> Eshell V5.5.4  (abort with ^G)
It seems a problem with the the latest version of erlang.

I have two versions of erlang installed:
- v5.5.5 (R11B-5) compiled from sources and here I still get the same error
- v5.5.2 installed with 'apt-get erlang' works as expected.

I have an AMD Turion 64x2 laptop running ubuntu 7.04

> 
> I will make some slight changes, e.g. make sure that you don't
> have to restart after having run erlhive:run_this_once()
> (Actually, all that's needed is that you call erlhive:start()).
This already seems to work. In my previous post I was following the
USER_GUIDE but it's not easy to keep documentation up to date with
project changes :-)

> 
> I also noticed that the file system support is broken. ):
> I'll take a look at that.
> 
> (Apologies for the printouts, but perhaps they can be of some
> pedagogical use in the beginning? I will remove them eventually.)
> 
Yes they are really useful to better understand how things work.

BR,
filippo

Re: [Erlhive-devel] problem getting started

[Ulf W. <ul...@wi...>]

Hi Filippo,

I just tried with the latest version (trunk):

Eshell V5.5.4  (abort with ^G)
1> erlhive:run_this_once().
put_trigger(new_account, <<"root">>) Pid = <0.91.0>
put_trigger(new_account, <<"otp">>) Pid = <0.91.0>
run_triggers([{new_account,<<"root">>},{new_account,<<"otp">>}],
<<"root">>) Pid = <0.31.0>
install({erlhive_user,{acct,<<"root">>,
                            user,
                            <<"root">>,
                            {{2007,9,13},{14,3,8}},
                            root_meta,
                            root,
                            0,
                            100,
                            internal,
                            false}}) = true
install({erlhive_user,{acct,<<"otp">>,
                            user,
                            <<"otp">>,
                            {{2007,9,13},{14,3,8}},
                            otp_meta,
                            otp,
                            0,
                            100,
                            internal,
                            false}}) = true
true
2> erlhive:start().
waiting for core tables : [erlhive_accounts,erlhive_code,erlhive_modnames]...
waiting for user tables: [otp_meta,otp,root_meta,root]...
all loaded.
rebuilding...
rebuilding accounts -> {module,erlhive_acct_lookup}
ok
3> erlhive:admin(fun(M) -> M:list_users() end).
run_triggers([], <<"root">>) Pid = <0.31.0>
[<<"guest">>,<<"otp">>,<<"root">>]


I will make some slight changes, e.g. make sure that you don't
have to restart after having run erlhive:run_this_once()
(Actually, all that's needed is that you call erlhive:start()).

I also noticed that the file system support is broken. ):
I'll take a look at that.

(Apologies for the printouts, but perhaps they can be of some
pedagogical use in the beginning? I will remove them eventually.)

BR,
Ulf W

2007/9/13, Filippo Pacini <pa...@sg...>:
> Hi,
> I'm an erlhive newbie. I'm trying it from the shell following the
> USER_GUIDE.
>
> After creating the db and restarting the shell and mnesia,
> trying to run admin commands I always get a badfun error. E.g.:
>
> > erlhive:admin(fun(M) -> M:list_users() end).
>
> ** exited: {{aborted,{{badfun,{{erlhive_admin},list_users}},
>                       [{shell,apply_fun,3},
>                        {erlhive_watchdog,with_watchdog,1},
>                        {mnesia_tm,apply_fun,3},
>                        {mnesia_tm,execute_transaction,5},
>                        {mnesia_schema,schema_coordinator,3}]}},
>             [{erlhive,schema_activity,1},
>              {erl_eval,do_apply,5},
>              {shell,exprs,6},
>              {shell,eval_loop,3}]} **
>
> The same happens with create_user.
>
> I've found also a erlhive:start/0 but even if I run this before the
> erlhive:admin functions nothing changes.
>
> What am I doing wrong?
>
> cheers,
> filippo
>
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2005.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Erlhive-devel mailing list
> Erl...@li...
> https://lists.sourceforge.net/lists/listinfo/erlhive-devel
>

[Erlhive-devel] problem getting started

[Filippo P. <pa...@sg...>]

Hi,
I'm an erlhive newbie. I'm trying it from the shell following the
USER_GUIDE.

After creating the db and restarting the shell and mnesia,
trying to run admin commands I always get a badfun error. E.g.:

> erlhive:admin(fun(M) -> M:list_users() end).

** exited: {{aborted,{{badfun,{{erlhive_admin},list_users}},
                      [{shell,apply_fun,3},
                       {erlhive_watchdog,with_watchdog,1},
                       {mnesia_tm,apply_fun,3},
                       {mnesia_tm,execute_transaction,5},
                       {mnesia_schema,schema_coordinator,3}]}},
            [{erlhive,schema_activity,1},
             {erl_eval,do_apply,5},
             {shell,exprs,6},
             {shell,eval_loop,3}]} **

The same happens with create_user.

I've found also a erlhive:start/0 but even if I run this before the
erlhive:admin functions nothing changes.

What am I doing wrong?

cheers,
filippo

[Erlhive-devel] Fwd: SF.net SVN: erlhive: [61] trunk/lib/erlhive/src/otp/erlhive_prim_file.erl

[Ulf W. <ul...@wi...>]

All small change to trunk, but an interface change (not that this
particular interface was documented...)

When using the file module in erlhive, the files are actually stored
in mnesia, and committed with transaction semantics. Files under
/tmp/ are removed when the transaction terminates and other users'
public files can be accessed via /home/User/... (used to be /erlhive/User/...).
This last detail was the interface change.

(Let me know when I should start really worry about backward compatibility. ;-)

BR,
Ulf W

---------- Forwarded message ----------
From: uw...@us... <uw...@us...>
Date: 2007-sep-13 08:11
Subject: SF.net SVN: erlhive: [61]
trunk/lib/erlhive/src/otp/erlhive_prim_file.erl
To: ul...@wi...


Revision: 61
          http://erlhive.svn.sourceforge.net/erlhive/?rev=61&view=rev
Author:   uwiger
Date:     2007-09-12 23:11:33 -0700 (Wed, 12 Sep 2007)

Log Message:
-----------
changed /erlhive/ to /home/

Modified Paths:
--------------
    trunk/lib/erlhive/src/otp/erlhive_prim_file.erl

Modified: trunk/lib/erlhive/src/otp/erlhive_prim_file.erl
===================================================================
--- trunk/lib/erlhive/src/otp/erlhive_prim_file.erl     2007-09-07
14:09:35 UTC (rev 60)
+++ trunk/lib/erlhive/src/otp/erlhive_prim_file.erl     2007-09-13
06:11:33 UTC (rev 61)
@@ -1124,7 +1124,7 @@
            throw({error,einval})
     end.

-i_get_user(["/","erlhive",U|_]) ->
+i_get_user(["/","home",U|_]) ->
     list_to_binary(U);
 i_get_user(["/", "tmp" | _]) ->
     Acct#acct.name;


This was sent by the SourceForge.net collaborative development
platform, the world's largest Open Source development site.

[Erlhive-devel] newest stuff in trunk

[Ulf W. <ul...@wi...>]

I've finally committed my latest stuff to trunk

svn co http://erlhive.svn.sourceforge.net/svnroot/erlhive/trunk

Perhaps the biggest change is that I've started building on
a erlhive/src/otp directory, currently with the following contents:

erlhive_compile.erl
erlhive_epp.erl
erlhive_file.erl
erlhive_file_io_server.erl
erlhive_file_server.erl
erlhive_ftp.erl
erlhive_gen.erl
erlhive_gen_server.erl
erlhive_gen_tcp.erl
erlhive_io.erl
erlhive_os.erl
erlhive_prim_file.erl
erlhive_proc_lib.erl

The file and io stuff works, practically all of it.
You can read other users' files, by addressing them
as /erlhive/User/... This is akin to reading public data from
another user. Anything you put under /tmp will be removed
when the transaction terminates. (:

I've not yet quite figured out erlhive_compile, so it's just
a copy of the original compile.erl. My idea is that you should
be able to call compile:file(...) from the erlhive shell, and have
erlhive automatically compile, transform, and output beam
files to the "erlhive file system". The FTP support will work
for fetching files from another FTP server. This could be used
as a convenient way to bring code into your erlhive account.

epp works in erlhive, though, FWIW. I have this hope that I will
be able to move over entirely to file-based compile and load,
but I have to figure out how to best avoid subversion (loading
a module that hasn't been safe-compiled). The answer is
probably attributes - since all files in erlhive are actually
database objects, you can attach attributes to them, and
erlhive has a set of "automatic" attributes that can't be
modified by user code.

I will not swear that all of this works right now. I've been a bit
too busy to play with it lately.

Oh, and I put a link to the Erlhive Wiki on the web page.

BR,
Ulf W

Re: [Erlhive-devel] major reorg

[Martin L. <mar...@gm...>]

On 8/20/07, Ulf Wiger <ul...@wi...> wrote:
> The svn manuals should warn clearly against creating repositories
> without a trunk/ directory. While it does say that it's a good idea,
> it could well emphasize just how bad an idea it is _not_ to. (:


I was bit by the same thing when I started using SVN.  It is too bad
that proper structure is just an idiom and not enforced by SVN.

>
>
> I've decided to try to reorganize the erlhive code tree,
> adding trunk/ and branches/ directories, and moving
> all currently checked in code under trunk/ (using svn mv).
>
> The structure should then look like this:
>
>   trunk/
>    +- proj_web/
>    +- lib/
>      +- erlhive/
>      +- nofrills/
>   branches/
>
> Then I thought I'd copy everything under trunk to branches/R1A,
> and then commit my new stuff.
>
> Any seasoned svn experts who feel this is madness, please give me a shout.
>
> Anyone with tons of code checked out, let me know.
>
> Otherwise, I'll just commit this, and then start moving my new
> stuff into trunk/
>
> BR,
> Ulf W
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >>  http://get.splunk.com/
> _______________________________________________
> Erlhive-devel mailing list
> Erl...@li...
> https://lists.sourceforge.net/lists/listinfo/erlhive-devel
>

Re: [Erlhive-devel] major reorg

[Ulf W. <ul...@wi...>]

No objections were voiced, so I went ahead with the reorg.
I've now created a beta-1.0 branch which contains the old
code. I tried running the test code, and nofrills also works,
after some slight changes in the Makefile and yaws.conf.src.

It might take some time to merge my new code into trunk.

BR,
Ulf W

2007/8/20, Ulf Wiger <ul...@wi...>:
> The svn manuals should warn clearly against creating repositories
> without a trunk/ directory. While it does say that it's a good idea,
> it could well emphasize just how bad an idea it is _not_ to. (:
>
>
> I've decided to try to reorganize the erlhive code tree,
> adding trunk/ and branches/ directories, and moving
> all currently checked in code under trunk/ (using svn mv).
>
> The structure should then look like this:
>
>   trunk/
>    +- proj_web/
>    +- lib/
>      +- erlhive/
>      +- nofrills/
>   branches/
>
> Then I thought I'd copy everything under trunk to branches/R1A,
> and then commit my new stuff.
>
> Any seasoned svn experts who feel this is madness, please give me a shout.
>
> Anyone with tons of code checked out, let me know.
>
> Otherwise, I'll just commit this, and then start moving my new
> stuff into trunk/
>
> BR,
> Ulf W
>

Re: [Erlhive-devel] erlhive blog_test

[yin s. <sun...@gm...>]

I didn't try blog_test. But the one under nofrills-1.0/src/ is working
(although it is just a demo, not fully functioning)
what I did is cd nofrills-1.0/src/
start erlang shell, make sure erlhive is in your path
> c(blog).
> blog:install_db().
after finish,  shell will be auto-exited.
start yaws, make sure it uses $ERLHIVE/nofrills-1.0/yaws.conf and erlhive is
in the path (-pa $ERLHIVE/lib/erlhive/ebin)
ofcourse, default database is the one your just created at nofrills-1.0/src/
> mnesia:start().

now you can use your web browser to access the blog interface at
nofrills-1.0/site/run.html

/Yin



On 8/19/07, Bernhard Damberger <be...@gm...> wrote:
>
> I was trying to get erlhive to run on my machine w/ the blog_test
> example. However, I am running into the following failure:
>
> Eshell V5.5.5  (abort with ^G)
> 1> blog_test:run_once().
> ** exited: {aborted,{{invalid_op,[{'___Env',0}]},
>                      [{erlhive_eval,local_fun_1,3},
>                       {erlhive_eval,local_fun,3},
>                       {erl_eval,local_func,5},
>                       {erl_eval,expr_list,6},
>                       {erl_eval,expr,5},
>                       {erl_eval,expr_list,6},
>                       {erl_eval,expr,5},
>                       {erl_eval,expr_list,6}]}} **
>
> If I start over (remove Mnesia dir) and comment out the line
> {init_blog, init()}].
> in run_once/0 I get the following:
>
> Eshell V5.5.5  (abort with ^G)
> 1> blog_test:run_once().
> [{start_erlhive,true},
> {add_users,[<<"blog">>,<<"guest">>,<<"root">>,<<"user1">>,<<"user2">>]},
> {define_mods,true},
> {load_mods,[{ok,[{".",[{0,erl_lint,{unused_function,{'___Env',0}}}]}]},
>              {ok,[]}]}]
>
> If I run blog_test:load_mods/0 I get:
>
> 2> blog_test:load_mods().
> [{ok,[{".",[{0,erl_lint,{unused_function,{'___Env',0}}}]}]},{ok,[]}]
>
> Looks like this corresponds to flexiblog.erl somehow.
> Not sure what all this means as I am new to erlhive (and erlang).
>
> Any pointers appreciated.
>
> _bernhard
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >>  http://get.splunk.com/
> _______________________________________________
> Erlhive-devel mailing list
> Erl...@li...
> https://lists.sourceforge.net/lists/listinfo/erlhive-devel
>

[Erlhive-devel] major reorg

[Ulf W. <ul...@wi...>]

The svn manuals should warn clearly against creating repositories
without a trunk/ directory. While it does say that it's a good idea,
it could well emphasize just how bad an idea it is _not_ to. (:


I've decided to try to reorganize the erlhive code tree,
adding trunk/ and branches/ directories, and moving
all currently checked in code under trunk/ (using svn mv).

The structure should then look like this:

  trunk/
   +- proj_web/
   +- lib/
     +- erlhive/
     +- nofrills/
  branches/

Then I thought I'd copy everything under trunk to branches/R1A,
and then commit my new stuff.

Any seasoned svn experts who feel this is madness, please give me a shout.

Anyone with tons of code checked out, let me know.

Otherwise, I'll just commit this, and then start moving my new
stuff into trunk/

BR,
Ulf W

Re: [Erlhive-devel] erlhive blog_test

[Ulf W. <ul...@wi...>]

I realize I didn't quite finish the sentence:

2007/8/20, Ulf Wiger <ul...@wi...>:
> 2007/8/20, Bernhard Damberger <be...@gm...>:
> > Any ideas on when your next commit might be?
>
> I figured I would copy the current version to a branch,
> and then commit my latest changes. It shouldn't take
> more than a few hours to make sure that it gets done
> properly (I'm not that experienced with svn).

... once I get a few hours to spare. I can't say exactly
when that will be, but I'll do my best to make it soon.

BR,
Ulf W

Re: [Erlhive-devel] erlhive blog_test

[Ulf W. <ul...@wi...>]

2007/8/20, Bernhard Damberger <be...@gm...>:
> Any ideas on when your next commit might be?

I figured I would copy the current version to a branch,
and then commit my latest changes. It shouldn't take
more than a few hours to make sure that it gets done
properly (I'm not that experienced with svn).


> I have to say that erlhive intrigues me greatly. But wrapping my brain
> around it seems to be taking a bit. Would you say the current version is
> good enough for writing any apps w/, or would you suggesting waiting
> until your next commit?

I have tried to keep the new stuff compatible with the old stuff.
Most of the changes have been about making it possible to
write apps that look more like 'normal erlang', and breaking the
erlhive code up into more manageable modules.

Joe's authenticating front end (the one in the nofrills-1.0 directory) is
obviously not ready for serious apps, partly because it has a habit of
revealing the passwords in plain text (just so you won't forget).
Some work needs to be done here, either extracting relevant code
and integrating it into a more conventional yaws framework, or
completing the existing front-end.

BR,
Ulf W

Re: [Erlhive-devel] erlhive blog_test

[Bernhard D. <be...@gm...>]

Any ideas on when your next commit might be?

I have to say that erlhive intrigues me greatly. But wrapping my brain
around it seems to be taking a bit. Would you say the current version is
good enough for writing any apps w/, or would you suggesting waiting until
your next commit?

_bernhard

On 8/19/07, Ulf Wiger <ul...@wi...> wrote:
>
> The warning is nothing to worry about, but the crash obviously is.
> To be honest, I don't think I've tested the blog_test code in ages.
> I'll try to take a look at it as soon as possible.
>
> (Most likely, I need to commit my latest stuff. This looks like
> an old bug in the code rewrite module.)
>
> BR,
> Ulf W
>
>
> 2007/8/19, Bernhard Damberger <be...@gm...>:
> > I was trying to get erlhive to run on my machine w/ the blog_test
> > example. However, I am running into the following failure:
> >
> > Eshell V5.5.5  (abort with ^G)
> > 1> blog_test:run_once().
> > ** exited: {aborted,{{invalid_op,[{'___Env',0}]},
> >                      [{erlhive_eval,local_fun_1,3},
> >                       {erlhive_eval,local_fun,3},
> >                       {erl_eval,local_func,5},
> >                       {erl_eval,expr_list,6},
> >                       {erl_eval,expr,5},
> >                       {erl_eval,expr_list,6},
> >                       {erl_eval,expr,5},
> >                       {erl_eval,expr_list,6}]}} **
> >
> > If I start over (remove Mnesia dir) and comment out the line
> > {init_blog, init()}].
> > in run_once/0 I get the following:
> >
> > Eshell V5.5.5  (abort with ^G)
> > 1> blog_test:run_once().
> > [{start_erlhive,true},
>
> >  {add_users,[<<"blog">>,<<"guest">>,<<"root">>,<<"user1">>,<<"user2">>]},
> >  {define_mods,true},
> >  {load_mods,[{ok,[{".",[{0,erl_lint,{unused_function,{'___Env',0}}}]}]},
> >              {ok,[]}]}]
> >
> > If I run blog_test:load_mods/0 I get:
> >
> > 2> blog_test:load_mods().
> > [{ok,[{".",[{0,erl_lint,{unused_function,{'___Env',0}}}]}]},{ok,[]}]
> >
> > Looks like this corresponds to flexiblog.erl somehow.
> > Not sure what all this means as I am new to erlhive (and erlang).
> >
> > Any pointers appreciated.
> >
> > _bernhard
> >
> >
> -------------------------------------------------------------------------
> > This SF.net email is sponsored by: Splunk Inc.
> > Still grepping through log files to find problems?  Stop.
> > Now Search log events and configuration files using AJAX and a browser.
> > Download your FREE copy of Splunk now >>  http://get.splunk.com/
> > _______________________________________________
> > Erlhive-devel mailing list
> > Erl...@li...
> > https://lists.sourceforge.net/lists/listinfo/erlhive-devel
> >
>

Re: [Erlhive-devel] erlhive blog_test

[Ulf W. <ul...@wi...>]

The warning is nothing to worry about, but the crash obviously is.
To be honest, I don't think I've tested the blog_test code in ages.
I'll try to take a look at it as soon as possible.

(Most likely, I need to commit my latest stuff. This looks like
an old bug in the code rewrite module.)

BR,
Ulf W


2007/8/19, Bernhard Damberger <be...@gm...>:
> I was trying to get erlhive to run on my machine w/ the blog_test
> example. However, I am running into the following failure:
>
> Eshell V5.5.5  (abort with ^G)
> 1> blog_test:run_once().
> ** exited: {aborted,{{invalid_op,[{'___Env',0}]},
>                      [{erlhive_eval,local_fun_1,3},
>                       {erlhive_eval,local_fun,3},
>                       {erl_eval,local_func,5},
>                       {erl_eval,expr_list,6},
>                       {erl_eval,expr,5},
>                       {erl_eval,expr_list,6},
>                       {erl_eval,expr,5},
>                       {erl_eval,expr_list,6}]}} **
>
> If I start over (remove Mnesia dir) and comment out the line
> {init_blog, init()}].
> in run_once/0 I get the following:
>
> Eshell V5.5.5  (abort with ^G)
> 1> blog_test:run_once().
> [{start_erlhive,true},
>  {add_users,[<<"blog">>,<<"guest">>,<<"root">>,<<"user1">>,<<"user2">>]},
>  {define_mods,true},
>  {load_mods,[{ok,[{".",[{0,erl_lint,{unused_function,{'___Env',0}}}]}]},
>              {ok,[]}]}]
>
> If I run blog_test:load_mods/0 I get:
>
> 2> blog_test:load_mods().
> [{ok,[{".",[{0,erl_lint,{unused_function,{'___Env',0}}}]}]},{ok,[]}]
>
> Looks like this corresponds to flexiblog.erl somehow.
> Not sure what all this means as I am new to erlhive (and erlang).
>
> Any pointers appreciated.
>
> _bernhard
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >>  http://get.splunk.com/
> _______________________________________________
> Erlhive-devel mailing list
> Erl...@li...
> https://lists.sourceforge.net/lists/listinfo/erlhive-devel
>

[Erlhive-devel] erlhive blog_test

[Bernhard D. <be...@gm...>]

I was trying to get erlhive to run on my machine w/ the blog_test
example. However, I am running into the following failure:

Eshell V5.5.5  (abort with ^G)
1> blog_test:run_once().
** exited: {aborted,{{invalid_op,[{'___Env',0}]},
                     [{erlhive_eval,local_fun_1,3},
                      {erlhive_eval,local_fun,3},
                      {erl_eval,local_func,5},
                      {erl_eval,expr_list,6},
                      {erl_eval,expr,5},
                      {erl_eval,expr_list,6},
                      {erl_eval,expr,5},
                      {erl_eval,expr_list,6}]}} **

If I start over (remove Mnesia dir) and comment out the line
{init_blog, init()}].
in run_once/0 I get the following:

Eshell V5.5.5  (abort with ^G)
1> blog_test:run_once().
[{start_erlhive,true},
 {add_users,[<<"blog">>,<<"guest">>,<<"root">>,<<"user1">>,<<"user2">>]},
 {define_mods,true},
 {load_mods,[{ok,[{".",[{0,erl_lint,{unused_function,{'___Env',0}}}]}]},
             {ok,[]}]}]

If I run blog_test:load_mods/0 I get:

2> blog_test:load_mods().
[{ok,[{".",[{0,erl_lint,{unused_function,{'___Env',0}}}]}]},{ok,[]}]

Looks like this corresponds to flexiblog.erl somehow.
Not sure what all this means as I am new to erlhive (and erlang).

Any pointers appreciated.

_bernhard

Re: [Erlhive-devel] is erlhive still alive ?

[Ulf W. <ul...@wi...>]

Hi Roberto,

I've not given up on erlhive just yet. ;-)

It is true that there is a major commit in the works. I thought
I'd hone my svn skills a bit, and create a trunk branch for it.
But so far, it should be compatible with the existing stuff.

I had hoped to be able to do this during the Summer holiday,
but a series of unplanned events made that impossible.

One of the things I've been working on is indeed making
erlhive more modular. A lot of the effort has gone into making
erlhive support as much as possible of regular OTP, e.g. the
file module, gen_server, gen_tcp, etc., but another way of
making apps erlhive-friendly is of course exporting some of
the patterns, so that they can be used even outside of erlhive.

I will take a closer look at that.

BR,
Ulf W

2007/7/28, Roberto Saccon <rs...@gm...>:
> I played with erlhive when it was announced and really liked the ideas
> behind it. Now it seems to be quiet on this list. Ulf, is development
> still going on ? Is there major commit to be expected at near future,
> which will turn everything existing irrelevant ?
>
> Now that I am a few months more experienced with Erlang, I am
> interested in integrating some of the features of erlhive  in an app
> of mine, e.g. the virtual mnesia based filesystem. Can it be decoupled
> from erlhive (I mean without too much problems) ?
>
> regards
> --
> Roberto Saccon
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >>  http://get.splunk.com/
> _______________________________________________
> Erlhive-devel mailing list
> Erl...@li...
> https://lists.sourceforge.net/lists/listinfo/erlhive-devel
>

[Erlhive-devel] is erlhive still alive ?

[Roberto S. <rs...@gm...>]

I played with erlhive when it was announced and really liked the ideas
behind it. Now it seems to be quiet on this list. Ulf, is development
still going on ? Is there major commit to be expected at near future,
which will turn everything existing irrelevant ?

Now that I am a few months more experienced with Erlang, I am
interested in integrating some of the features of erlhive  in an app
of mine, e.g. the virtual mnesia based filesystem. Can it be decoupled
from erlhive (I mean without too much problems) ?

regards
-- 
Roberto Saccon

[Erlhive-devel] Erlhive wiki

[Ulf W. <ul...@wi...>]

In the early days of Erlhive (the *really early* days), Joe created a Google
Code project and I created a Sourceforge project. The code ended up in the
Sourceforge project, but the Google code project is still around.

I've been thinking that there should be an Erlhive Wiki for documentation
and tutorials.

Well, the Google Code project has a wiki:

http://code.google.com/p/erlhive/wiki/ErlhiveIntro


Not that I'll  have much time at the moment to put stuff here right now, but

I thought I'd advertise its existence. I did insert the old User's Guide, so
there'll
at least be something there.

BR,
Ulf W

Re: [Erlhive-devel] stringtemplate

[Ulf W. <ul...@wi...>]

> A good solution IMHO is to define a template interface so
> that different template engines can easily be integrated.

That's surely worth considering.

I have some priority work in front of me for a while longer,
but will then wrap up my latest changes and commit them.

Apologies for the low intensity on the erlhive front, but
there's not much I can do about it at the moment.
Things are pretty hectic on other fronts. (:

(Which reminds me: I have to get back to my work...)

BR,
Ulf W

2007/5/27, Filippo Pacini <pa...@sg...>:
>
> Hi all,
> I'm the author of sgte. If you decide to start using it in erlhive and
> need help porting it let me know.
> I was planning to try using sgte in erlhive sooner or later, but I have
> not much time at the moment.
>
> Ulf Wiger wrote:
> > [cut]
> > (Not that there isn't room for many different template systems in
> > Erlhive, but I'd like to pick one for writing admin pages and examples,
> > and eventually start collecting a bunch of template designs.)
>
> A good solution IMHO is to define a template interface so that different
> template engines can easily be integrated.
>
> Take a look at this post from the Yaws mailing-list:
> http://www.nabble.com/sgte---erlang-template-engine-t3385899.html#a9431195
>
> cheers,
> filippo
>
>

Re: [Erlhive-devel] stringtemplate

[Filippo P. <pa...@sg...>]

Hi all,
I'm the author of sgte. If you decide to start using it in erlhive and
need help porting it let me know.
I was planning to try using sgte in erlhive sooner or later, but I have
not much time at the moment.

Ulf Wiger wrote:
> [cut] 
> (Not that there isn't room for many different template systems in
> Erlhive, but I'd like to pick one for writing admin pages and examples,
> and eventually start collecting a bunch of template designs.)

A good solution IMHO is to define a template interface so that different
template engines can easily be integrated.

Take a look at this post from the Yaws mailing-list:
http://www.nabble.com/sgte---erlang-template-engine-t3385899.html#a9431195

cheers,
filippo

[Erlhive-devel] stringtemplate

[Ulf W. <ul...@wi...>]

To open up the HTML template question again, I'm currently considering
just using SGTE (http://code.google.com/p/sgte/), the Erlang version of
StringTemplate, for erlhive.

While I've come a fairly long way towards supporting ErlyWeb (I'll
eventually
commit the new stuff...), I've started doubting whether importing a
scaffolding
framework into Erlhive (which is sort of a framework by itself) is really
the best
way forward.

I like the stringtemplate approach, and I haven't seen anything in the
SGTE code that wouldn't compile without changes in Erlhive.

(Not that there isn't room for many different template systems in
Erlhive, but I'd like to pick one for writing admin pages and examples,
and eventually start collecting a bunch of template designs.)

Comments?

BR,
Ulf W

[Erlhive-devel] Fwd: a plea for cleaner modules

[Ulf W. <ul...@wi...>]

Haven't posted in a while... forgot the address to my own list.

---------- Forwarded message ----------
From: Ulf Wiger <ul...@wi...>
Date: 2007-maj-09 21:35
Subject: a plea for cleaner modules
To: erlang-questions <erl...@er...>
Cc: erl...@so...


I spent some time on the train hacking away at Erlhive.
Now that I'm able to safely handle processes, ets tables
and files (at least virtual files) inside mnesia transactions,
I thought I'd throw in a HTTP or FTP client as well.

I was disappointed to find that neither  the ftp nor the http
module are especially clean, doing just what you'd expect.

My first step in checking how much work is needed to port
something into erlhive goes something like this:

2> Beam = code:which(ftp).
"c:/Program/ERL55~1.4/lib/inets-4.7.11/ebin/ftp.beam"
4> {ok,{_,[{imports,Is}]}} = beam_lib:chunks(Beam,[imports]).
{ok,{ftp,[{imports,[{application,start,1},
                    {dbg,p,2},
                   ...]}}
5> Ms = ordsets:from_list([M || {M,_,_} <- Is]).
[application,
 dbg,
 erlang,
 error_logger,
 file,
 filename,
 ftp_progress,
 ftp_response,
 ftp_sup,
 gen_server,
 gen_tcp,
 inet,
 inet_db,
 io_lib,
 lists,
 string]

Now, going through the list is a bit discouraging.
The following dependencies would cause porting problems:

  [application, dbg, error_logger]

(Ok, inet_db needs to be handled too, but that's a logical step.)

None of these are needed for the ftp logic, and so I'm forced
to first cut out portions of the code. After that, it should be a
pretty clean recompile.

Now, the bulk of ftp.erl is perfectly fine. It is, as it should be,
just a gen_server (but why on earth is it hard-coded to run
at low priority?!). But making the module double as a supervisor
behaviour, and automatically bootstrapping the inets application,
is unnecessary, I think.

Why not have a clean ftp gen_server module, and put all the
gunk in a wrapper module for those who want it?

Even if porting to Erlhive isn't high on your list, I can think of
many good reasons to write the main logic in a separate
module, with as few dependencies as possible.

BR,
Ulf W

[Erlhive-devel] new version - major, major rewrite

[Ulf W. <ul...@wi...>]

Those of you who subscribe to SVN commit messages have just received a  
giant mail.

I've done some major refactoring and added lots of functionality.

I don't think I have broken anything, but please let me know.

The best way to see some of the new stuff is to look at the
new basic examples (ex7 and on). Here's a list of highlights:

- Macros now work, as long as they are syntactically'
   complete.

- -import/1 and -import/2 work, which should make things
   easier.

- Processes (!). You can spawn processes and send and receive
   messages, but visibility is limited to the current
   transaction (no permanent processes allowed).

- Ets tables were there before, right? They still work.

I've split the code into more managable modules. This should
make it a lot easier to work with and maintain. The source
transformation has also been simplified (I think).

I'm also working on a new 'example' application, called 'otp'.
I have an erlhive version of file.erl, and it provides some
basic file handling services (files are stored as binaries
in Mnesia, and full commit/rollback semantics apply).

I've also started trying to compile as many modules as possible
 from stdlib et al. I've had success with gen.erl, gen_server.erl,
proc_lib.erl, epp.erl, ... but I haven't tested them yet.

The real test is to port the io subsystem over. The main use
would be for file handling, compiling, etc. I'm toying with the
idea of giving each user a "file system", where e.g. lib/ is
for code, and (hopefully, eventually) compiling etc work as
expected in an OTP system. Hopefully, most of this can be
done on top of the basic erlhive system.

(I have edited the web page, but haven't published it.
It's been so long, I'll have to read up on how it's done. ;)

Feedback, please.

BR,
Ulf W
-- 
Ulf Wiger

[Erlhive-devel] ping - almost there

[Ulf W. <ul...@wi...>]

I just thought I'd issue a life sign.

I believe that I have managed to add process support to
erlhive, and am now in the process of compiling various
modules in stdlib to verify that it works. At the moment,
it doesn't quite work in the tiny details, but progress
is good (the few hours I'm able to work on it).

What I've done is basically to allow spawning of processes,
message passing, linking, etc., but only within the
context of a transaction, with almost the same visibility
rules as for data. If a nested transaction is started,
any processes in that nested transaction will see only
each other, and will be killed (yes, killed) when the
transaction aborts or commits.

This is of course limiting, but it's pretty easy to
reason about. Comments are most welcome.

As a reminder, I want to make it as easy as possible
to port code into erlhive, and processes and ets
tables are needed for many interesting modules, e.g.
the io subsystem, epp, erlyweb, ...

BR,
Ulf W
-- 
Ulf Wiger

[Erlhive-devel] minor problems remain

[Ulf W. <ul...@wi...>]

My virtual scheduler seems to work, at least in a simple example with two  
processes.

Unfortunately, one of my other examples crashed, since my use of
erl_syntax_lib:annotate_bindings/2 didn't work for try clauses
(not yet implemented in erl_syntax_lib).

I haven't checked in anything yet.


The function doesn't seem awfully difficult to write, but it's
always a pain to have to mandate local patches of OTP code.

I also need to see whether my new constructs actually work
for porting legacy code into erlhive.

(If I learned some more about subversion, I'm sure I could
check in this stuff in a separate branch - ideas, anyone?)

BR,
Ulf W
-- 
Ulf Wiger