Forum: Ruby retry / redo question

Announcement (2017-05-07): www.ruby-forum.com is now read-only since I unfortunately do not have the time to support and maintain the forum any more. Please see rubyonrails.org/community and ruby-lang.org/en/community for other Rails- und Ruby-related community platforms.
C6ec9468bab19475922dd29d00714da8?d=identicon&s=25 Rebhan, Gilbert (Guest)
on 2007-02-26 13:41
(Received via mailing list)
Hi,

i have a ruby script that does for 1 - n  folders a
cp_r of every folder to an existing cvs workspace
and afterwards the cvs operations
update to determine if add is needed
add if needed 1-n times
commit

As the cvs CLI lacks a recursive parameter for
the update / add  commands, means when a new file
like subfolder/subfile.txt comes in
the add command only recognizes subfolder, but the
subfile.txt only after a 2nd add command

 i have to do =

1. updpipe=IO.popen("#{CVSEXE} -d #{CVSROOT} update")
     updpipe.readlines.each { |x|
       updfiles<<'.'<<'/'<<x[2..-2]<<' ' }

to determine whether there are new files to be added

if updfiles != 0 i have to execute the cvs add command =

2.addpipe=IO.popen("#{CVSEXE} -d #{CVSROOT} add #{updfiles.to_s}")
   addpipe.readlines.each { |x|
   addfiles<<'.'<<'/'<<x[2..-2]<<' ' }

this add command gives me other folders or files that are also to be
added, if there are more to be added, i.e subfolders/subfiles ...

so i have to do the cvs add in a loop until nothing more comes
back from that cvs add command.

then i have to do a final commit =

3.system("#{CVSEXE}","-d","#{CVSROOT}","commit","-m","bla bla bla bla")


Now i'm looking for the best loop construct, i know
the addfiles array has to be cleared after every cvs add before
executing cvs add again

There is something like retry/redo i tried with retry after and before
the end of a while loop

updfiles=Array.new
updpipe=IO.popen("#{CVSEXE} -d #{CVSROOT} update")
     updpipe.readlines.each { |x|
       updfiles<<'.'<<'/'<<x[2..-2]<<' ' }

while updfiles.length > 0
  addfiles=Array.new
  addpipe=IO.popen("#{CVSEXE} -d #{CVSROOT} add #{updfiles.to_s}")
  addpipe.readlines.each { |x|
  addfiles<<'.'<<'/'<<x[2..-2]<<' ' }
updfiles.clear
addfiles.clear
  updpipe.readlines.each { |x|
  updfiles<<'.'<<'/'<<x[2..-2]<<' ' }
end
*retry*

but got :

retry outside of rescue clause (LocalJumpError)


What's the right loop ?
Is there a solution without retry / redo ?


Regards, Gilbert
E0d864d9677f3c1482a20152b7cac0e2?d=identicon&s=25 Robert Klemme (Guest)
on 2007-02-26 16:05
(Received via mailing list)
On 26.02.2007 13:40, Rebhan, Gilbert wrote:
> the update / add  commands, means when a new file
> to determine whether there are new files to be added
> so i have to do the cvs add in a loop until nothing more comes
>
>   addpipe=IO.popen("#{CVSEXE} -d #{CVSROOT} add #{updfiles.to_s}")
>
> retry outside of rescue clause (LocalJumpError)
>
>
> What's the right loop ?
> Is there a solution without retry / redo ?

If you want to use retry you need to do that in an rescue clause as the
error message indicates.

However, I think, you should rather change the logic.

First, I'd start with using http://raa.ruby-lang.org/project/ruby-cvs/
to make CVS accesses more efficient (you do not start a process for each
CVS command).

Then, if you do the recursive traversal through the file tree (probably
using Find.find) I'd add a non existing directory immediately whenever I
see it.  That way you can be sure that all directories on the path are
present in the repository before you add individual files.  HTH

Regards

  robert
C6ec9468bab19475922dd29d00714da8?d=identicon&s=25 Rebhan, Gilbert (Guest)
on 2007-02-26 16:20
(Received via mailing list)
Hi,


/*
However, I think, you should rather change the logic.

First, I'd start with using http://raa.ruby-lang.org/project/ruby-cvs/
to make CVS accesses more efficient (you do not start a process for each

CVS command).
*/

I already thought of that too, need to have a detailled look,
whether this library works also with cvsnt, which we use here
as cvs server for windows.

i.e. cvsnt stores it's passwords in the registry, not in
user.home/.cvspass
etc.

/*
Then, if you do the recursive traversal through the file tree (probably
using Find.find)
I'd add a non existing directory immediately whenever I
see it.  That way you can be sure that all directories on the path are
present in the repository before you add individual files.
*/

i thought of something similar and tried with :

newfiles=Dir["Y:/test_deploy/**/*.*"] - Dir["Y:/cvsworkspace/**/*.*"]

but that gives me the whole path, i.e.
Y:/test_deploy/subfolder7/subsub7/bla.txt, whereas i need it as space
separated list for :
cvs add ./subfolder7 ./subfolder7/subsub7 ./subfolder7/subsub7/bla.txt

Any idea how this could get a working solution ?!


Regards, Gilbert
This topic is locked and can not be replied to.