Hi,
Once I have started installing Fedora 6 it works out the packages/dependencies and then waits for me to press the "Next" button before it starts the actual installtion.
Maybe it should just go ahead and do it without waiting for the user to press the "Next" button.
Is it possible to automate this already, if not then I think this will be a good feature to have?
Thanks Shams
After that point installation start writing to disk. Before that everything is happening without any change on the hdd.
It's a good pointer for user that after that moment s/he is starting to change his partitions. AFAIK, after this point anaconda do formatting of the partitions. One last click won't annoy user much, but losing whole partitions might.
However, I think this page should point this fact more explicitly. Stating that (if user chose to format any partitions) "After this part your data on 'blah blah' partitions will be cleared".
On Sun, 2007-03-11 at 02:27 +1300, Shams wrote:
Hi,
Once I have started installing Fedora 6 it works out the packages/dependencies and then waits for me to press the "Next" button before it starts the actual installtion.
Maybe it should just go ahead and do it without waiting for the user to press the "Next" button.
Is it possible to automate this already, if not then I think this will be a good feature to have?
Thanks Shams
Am Sonntag, den 11.03.2007, 18:08 +0200 schrieb Baris Cicek:
After that point installation start writing to disk. Before that everything is happening without any change on the hdd.
It's a good pointer for user that after that moment s/he is starting to change his partitions. AFAIK, after this point anaconda do formatting of the partitions. One last click won't annoy user much, but losing whole partitions might.
However, I think this page should point this fact more explicitly. Stating that (if user chose to format any partitions) "After this part your data on 'blah blah' partitions will be cleared".
You already get a warning before partitions are changed.
On 3/11/07, Baris Cicek baris@teamforce.name.tr wrote:
After that point installation start writing to disk. Before that everything is happening without any change on the hdd.
It's a good pointer for user that after that moment s/he is starting to change his partitions. AFAIK, after this point anaconda do formatting of the partitions. One last click won't annoy user much, but losing whole partitions might.
It would not annoy the user _IF_ the depsolving step was faster.
I think this is the reason why I saw mails in the past with the same request: a fairly long step (depsolving) followed by another long step (actual installation) sounds naturally like they should be merged.
IIRC the reason for this is that the depsolving phase could fail depending on the package selection and the selected repos; I still fail to see whay the last "Next" could not be skipped automatically when no such problem arise though...
On Sunday 11 March 2007 12:20pm, Gianluca Sforna wrote:
On 3/11/07, Baris Cicek baris@teamforce.name.tr wrote:
After that point installation start writing to disk. Before that everything is happening without any change on the hdd.
It's a good pointer for user that after that moment s/he is starting to change his partitions. AFAIK, after this point anaconda do formatting of the partitions. One last click won't annoy user much, but losing whole partitions might.
It would not annoy the user _IF_ the depsolving step was faster.
I think this is the reason why I saw mails in the past with the same request: a fairly long step (depsolving) followed by another long step (actual installation) sounds naturally like they should be merged.
IIRC the reason for this is that the depsolving phase could fail depending on the package selection and the selected repos; I still fail to see whay the last "Next" could not be skipped automatically when no such problem arise though...
In that case, the "After clicking 'Next' your partitions will be altered" thing will need to be mentioned before depsolving. If depsolve fails, then bounce back to the package select screen with a (hopefully) useful error message.
In other words, if this is going to be change, simply move depsolving to after the last confirm screen and right before partitioning, placing the final confirm question between package selection and depsolving. It certainly was logical to put package selection and depsolving together in the past, but perhaps this would be better.
Of course, if my suggestion of how to implement this change were adopted, then when depsolving does fail, experienced users/admins might be frustrated that they walked away thinking there wasn't going to be any more interaction and it should have installed when they come back and see an error message. Then again, that's always true.
On 3/11/07, Lamont Peterson lamont@gurulabs.com wrote:
On Sunday 11 March 2007 12:20pm, Gianluca Sforna wrote:
On 3/11/07, Baris Cicek baris@teamforce.name.tr wrote:
After that point installation start writing to disk. Before that everything is happening without any change on the hdd.
It's a good pointer for user that after that moment s/he is starting to change his partitions. AFAIK, after this point anaconda do formatting of the partitions. One last click won't annoy user much, but losing whole partitions might.
It would not annoy the user _IF_ the depsolving step was faster.
I think this is the reason why I saw mails in the past with the same request: a fairly long step (depsolving) followed by another long step (actual installation) sounds naturally like they should be merged.
IIRC the reason for this is that the depsolving phase could fail depending on the package selection and the selected repos; I still fail to see whay the last "Next" could not be skipped automatically when no such problem arise though...
In that case, the "After clicking 'Next' your partitions will be altered" thing will need to be mentioned before depsolving. If depsolve fails, then bounce back to the package select screen with a (hopefully) useful error message.
+1
Of course, if my suggestion of how to implement this change were adopted, then when depsolving does fail, experienced users/admins might be frustrated that they walked away thinking there wasn't going to be any more interaction and it should have installed when they come back and see an error message. Then again, that's always true.
If there is an error in the depsolving phase I don't see any difference from the current situation: you should go back and fix what went wrong.
The only difference is when no problems arise from depsolving, that is, no more interactions are needed
Hi,
Yes it is the depsolve that takes a long time, annoying really.
One way I think it could be solved is before depsolve have a checkbox, possibly intergrated into one of the existing screens:
"Start installing to disk automatically after solving dependecies after 60 (editable?) seconds?
which is checked OFF by default but since I know what I am doing I can check it on and hence in this case when the timeout happens the user does not have to click the "Next" button and install to disk happens automatically BUT is still able to cancel within the timeout period. In the case depsolve fails then the usual error handling happens.
Thanks Shams
On Mon, Mar 12, 2007 at 11:16:06 +1300, Shams shams@orcon.net.nz wrote:
Yes it is the depsolve that takes a long time, annoying really.
Has anyone looked into why it runs that slowly? Is it all of the disk seeks, doing the transitive closure or what? Is there enough information cached in comps.xml to compute the transitive closure without having to read all of the rpm files? Is the algorithm used to compute the transitive closure poor?
On Saturday 10 March 2007 06:27am, Shams wrote:
Hi,
Once I have started installing Fedora 6 it works out the packages/dependencies and then waits for me to press the "Next" button before it starts the actual installtion.
Maybe it should just go ahead and do it without waiting for the user to press the "Next" button.
Is it possible to automate this already,
I haven't tried this, but you could create a kickstart file with just one line that reads "install" and see if anaconda will take it. The "install" command is the one that tells anaconda to automatically say yes to that question.
If a kickstart file doesn't have all the answers, it asks just the questions it needs to. Say, for example, that you have a kickstart file without timezone information; anaconda will use the answers that are present for everything else, asking only the one question about timezone configuration.
if not then I think this will be a good feature to have?
Perhaps, perhaps not.
Yes I looked into it but want to be able to do it thru the gui:
Thanks Shams