Question about BuildArch and rpmrc on 64-bit

[Neal ManBear]

snake??? ... i have 15 snakes now lol ... and i am always quiet

     
Shhh!      

ok dad lol ::::::::::::::;flees:::::::::::::::::::::::::::::

     
You are in for it now.     
       

[travisN000]

Heya Travis,

Apart from the update which I now have installed and still have to be tested, I just like to point out a few things that might help improve the package.

On the update side, wouldn't it be better for pkgutils to check if there is already an exisiting .rpmmacros and .rpmrc, and have the option to review the changes if any? The same thing with mkrepo.

I do not know about the other packagers but I have an unconventional setup and do not have my /src folder right directly under my /home/archie. It is on a subfolder, which is the way I prefer.

It's not a big deal really but the updates prompt me to double check my setup and make sure that my configuration are still in place because I do not want my RPMS saved on other locations. I also noticed that mkrepo added the /home/archie/src URI and I already have one that I use and what was needed to be added was the i586, athlon and noarch section along with the x86_64.

Here is an updated version, renamed mkrepo-gui:  
http://dl.dropbox.com/u/2425248/mkrepo-gui
http://minus.com/me7TylqRY#5 (updated link)
(dropbox & minus links  )

..it should allow you to put the rpm build directories, local repo, and rpm build temp directories any place you like, and will also try to read existing values from your rpmmacros file!  

Please test thoroughly, as this turned out to be a fairly significant re-write of the code (..I did test, but more eyes are better).

If it seems ok i will update the pkgutils package again

[Archie]

Heya Travis,

First, thanks for making a GUI. Second, I'm getting this:

[img]http://i.minus.com/1322000657/1i-2VEPzW7CxEZiTBB060g/d7bCpHIoq73kn[img]

from clicking http://min.us/l7bCpHIoq73kn

        

[Neal ManBear]

The file does not exist or is unavailable

     

[Neal ManBear]

Got it now. I'll email it to you, Archie.     

[Archie]

Thanks, Neal.

[travisN000]

..by the way, the main dialog uses yad, not zenity like the rest, so yad is also required; I may still convert the rest of the dialogs to yad, but wanted to let people start testing the basic mechanics of the script.

Command line options are no longer supported in this version..  the reason being that the old version was written to not allow the use of directory names with spaces; it also was coded to have everything in a single directory tree.  

By allowing the user to pick directories in a GUI a much different set of error checking was needed and parts of it were contradictory with the old code.  If anyone is really attached to the console options I could restore the previous mkrepo to a console only script and include both, but the old one will not support set-ups like Archie's.

[Archie]

My first test went very well. And everything was according to what I wanted it to do except for one little thing. It created the /home/archie/src by itself. Of course it is no big deal and easily deleted but it shouldn't have, should it?

Screenshots: Minus

I'll test it on LiveUSB and a 32-bit install. Thanks again, travisN000.

[travisN000]

My first test went very well. And everything was according to what I wanted it to do except for one little thing. It created the /home/archie/src by itself. Of course it is no big deal and easily deleted but it shouldn't have, should it?

..it should delete those directories..  but in its current testing state I set it to just echo a message to the console; I didn't want it to delete anything unexpectedly. 
 
I did find one small issue...  directory names with a "_" character in them get converted to a space, and then throw an error because the script cannot write to them (..its looking for the wrong directory); I fixed it on my copy.    Directory names with a "|" (the character I chose as a delimiter in yad) will also likely throw an error, but I cannot fix this, as it is inherent to the way yad works.

[Archie]

My first test went very well. And everything was according to what I wanted it to do except for one little thing. It created the /home/archie/src by itself. Of course it is no big deal and easily deleted but it shouldn't have, should it?

..it should delete those directories..  but in its current testing state I set it to just echo a message to the console; I didn't want it to delete anything unexpectedly. 
 
I did find one small issue...  directory names with a "_" character in them get converted to a space, and then throw an error because the script cannot write to them (..its looking for the wrong directory); I fixed it on my copy.    Directory names with a "|" (the character I chose as a delimiter in yad) will also likely throw an error, but I cannot fix this, as it is inherent to the way yad works.

How about a check if the input contains any of those characters will return a try again with a notice that those characters are invalid in the directory names?

[travisN000]

I did find one small issue...  directory names with a "_" character in them get converted to a space, and then throw an error because the script cannot write to them (..its looking for the wrong directory); I fixed it on my copy.    Directory names with a "|" (the character I chose as a delimiter in yad) will also likely throw an error, but I cannot fix this, as it is inherent to the way yad works.

How about a check if the input contains any of those characters will return a try again with a notice that those characters are invalid in the directory names?

The issue with the underscore character (_) was solved by a slight change to how I handled spaces in directory names...  The possible issue with the "|" character cannot be checked for directly, because yad returns a single string of all field values, delimited by that character (..or any other I choose).  The delimiter character will always be present in the string.  

However, now that I think about it more, i don't think it will be an issue, as the check I run to make sure no field is left blank should also catch the situation with an extra delimiter character in the output (the script checks to make sure it recieves exactly 7 delimited items from the yad dialog).  I will just need to modify the error notification so that if a greater than expected number of delimiters is detected the error message is specific to this situation.  ..it still isn't a perfect solution though, because a combination of the two situations (an empty field and a field conataining the delimiter character) could still possibly fool the current logic into accepting bad input... it would split the field contianing the delimiter character into two fields and the result would be accepted by the logic I used.  I'll have to do some more testing on this over the weekend and see if I can find a way to catch both and still make sure the user gets correct feedback about the cause of the error.

[Archie]

I think I must have dozed off on "However, now that I think about it ..."

It shouldn't be a real issue ... I mean, who'd want to use a pipe on a directory name? But it's still good that you're getting all this exercise, travisN000 ... and I'm around to test.

Looking forward to the result.

[travisN000]

It shouldn't be a real issue ...

..I'm going with you on this one.  

I've uploaded a "final" version of just the script here: http://minus.com/me7TylqRY#2f

..it has a couple of minor changes to console output, and one bigger change.. ..it will now remove those directories created at the start-up of the script.  It does check to make sure they are still empty, so nothing should get deleted by accident!
 
If nobody finds any issues with this one, I will probably package it up and sent it to tex sometime tomorrow.


Also, if anyone is interested in testing a new version of apt-sources-update (aka repo-speed-test), I also posted a testing version of it here: http://www.pclinuxos.com/forum/index.php/topic,99755.msg849159.html#msg849159


Thanks again!  

[Archie]

It shouldn't be a real issue ...

..I'm going with you on this one.  

I've uploaded a "final" version of just the script here: http://minus.com/me7TylqRY#2f

..it has a couple of minor changes to console output, and one bigger change.. ..it will now remove those directories created at the start-up of the script.  It does check to make sure they are still empty, so nothing should get deleted by accident!
 
If nobody finds any issues with this one, I will probably package it up and sent it to tex sometime tomorrow.


Also, if anyone is interested in testing a new version of apt-sources-update (aka repo-speed-test), I also posted a testing version of it here: http://www.pclinuxos.com/forum/index.php/topic,99755.msg849159.html#msg849159


Thanks again!  

Going to test it on a LiveUSB setup. From the last version to this, I'm confident there are no other issues. Thank you very much, Travis. I am pretty sure your script will facilitate a sound setup for new (and current, if we have to reinstall and set everything up again) packagers.