Issue 7630 - Add -nogui option for some response installs (patch)
Summary: Add -nogui option for some response installs (patch)
Status: CLOSED FIXED
Alias: None
Product: Installation
Classification: Application
Component: code (show other issues)
Version: OOo 1.0.1
Hardware: PC Linux, all
: P3 Trivial (vote)
Target Milestone: ---
Assignee: falko.tesch
QA Contact: issues@installation
URL:
Keywords: oooqa
Depends on:
Blocks:
 
Reported: 2002-09-10 12:34 UTC by mmeeks
Modified: 2008-05-17 23:02 UTC (History)
3 users (show)

See Also:
Issue Type: ENHANCEMENT
Latest Confirmation in: ---
Developer Difficulty: ---


Attachments
patch to add -nogui (1.82 KB, patch)
2002-09-10 12:34 UTC, mmeeks
no flags Details | Diff

Note You need to log in before you can comment on or make changes to this issue.
Description mmeeks 2002-09-10 12:34:21 UTC
As you know it's pretty painful when building an OO package to have to
have a fake X server to satisfy the setup program; even when using a
response file.

        It seems this is a reasonable thing, since a response file is used for
2 purposes in the typical RPM setup:

        a) Creating the RPM
        b) Auto-setting up the user's home directory

        Clearly for the latter we need nice GUI dialogs where neccessary; for
the former we don't.

        Thus this (trivial) patch for setup2, vs. OOO_STABLE_1 adds the
'-nogui' option, which is used to explicitely disable the GUI in the
case that we're pushing a response file at setup2. This will help
cleanup some of the build mess for Unix packages, and contains very
little code.

        Any chance of having this on HEAD, SRX643 and OOO_STABLE_1 ?

        Then again - perhaps I want to be using a different 'response level' -
what are -rsp1 and -rsp2 good for ? they seem to not do 'much' for me
:-) I'm presumably missing some docs somewhere.

        Finally the code in setup2/mow/source/loader/loader.c [ which I'd like
to add a -debug conditional to inhibit the 'KillSetupDir' appears to
have a cut and paste block of code to parse the arguments.

        There are 2 possible problems; the MayUseX does case insensitive
checks, whereas that in 'main' does case insensitive ones (correctly),
and interestingly the loader differs from main.cxx inasmuch that it
doesn't initialize X with '-r:' whereas main.cxx does (for the
aforementioned reasons) - can someone clarify that for me ?
Comment 1 mmeeks 2002-09-10 12:34:56 UTC
Created attachment 2788 [details]
patch to add -nogui
Comment 2 Olaf Felka 2002-09-10 12:55:53 UTC
Hi Falko,
what do you think?
Comment 3 stx123 2002-09-11 13:23:52 UTC
I really like to have Falko's feedback, but could somebody from the 
installation project have a look at the patch in the meantime? If it 
works, why not include it?
Comment 4 dirk.voelzke 2002-09-13 13:51:50 UTC
I've not included the patch but found another solution. When we have a
response file, we only need vcl when we want to start the response
file autopilot ( -rspa ).
I've changed the parameter recognition to be case insensitive in
loader.c, too.
Last, in an src643 you can use the parameter -dontdeletetemp to
prevent the setup from removing the temp dir.
Comment 5 mmeeks 2002-09-13 14:41:33 UTC
I disagree;

We want to throw up GUI dialogs if a problem happens during the
first time auto-setup for a user of a network install.

Here is what typically happens:

a) build /install -net on a machine with no GUI with a response file

b) Install file set image

c) Run custom 'soffice' script that does a silent user install [ no
user wants to be bothered with filling in tens of lines of guff before
they can do anything / load a mail attachment the first time ]
   This process uses a different per-user response file doing a
minimal setup.

d) Run the user's installed version.

So - the problem with disabling the GUI is that if an error occurs
during step c) the GUI user will see nothing, no nice warning dialog,
no ability to change it.

Thus - it would be altogether better to apply the attached patch, and
revert the change disabling the GUI altogether for the response file
case - from my perspective at least.
Comment 6 stx123 2002-09-16 09:49:43 UTC
Michael, Dirk, would you like to follow-up on dev@installation or
continue in this issue? Greetings, Stefan.
Comment 7 dirk.voelzke 2002-09-16 10:12:18 UTC
Sorry, but the requirement has been 'no gui during response file
installation'. Therefor my fix is ok. Disabling the GUI with a
parameter is a nice idea, but you have to change much more than just
the parameter recognition. E.g. what should happen if somebody start
the setup with -nogui but without a responsefile? Perhaps we should do
it the other way and have an aditional parameter -showerrors? But
Stefan is right, we should continue on dev@installation.

Dirk
Comment 8 mmeeks 2002-09-16 13:47:41 UTC
My initial mail, explaining the issue in detail - was posted last
tuesday to dev@installation.openoffice.org: 6 days ago.

As yet - it has recieved precisely 0 replies; this doesn't fill  me
with confidence that anything will happen on-list. :-)
 
Strangely, my previous mail to that list also recieved no reply after
a week or so, leaving me to suspect the list is not used whatsoever
for any sort of development discussion.

I still tend to think the 'fix' is a mistake;
Comment 9 dirk.voelzke 2002-09-20 10:31:21 UTC
Although nobody should use GUI calls when using setup with
responsefiles, you can call hidesetup or showsetup which causes a core
dump when the gui was not initialized. Therefor i've rolled back my
changes and added the -nogui option. You will get an error message
when you try to use the -nogui option without a response file.
Comment 10 mmeeks 2002-09-20 10:44:44 UTC
Thanks that's great news.
Comment 11 dirk.voelzke 2003-04-03 12:31:52 UTC
*** Issue 12679 has been marked as a duplicate of this issue. ***
Comment 12 ace_dent 2008-05-17 21:00:12 UTC
The Issue you raised has been marked as 'Resolved' and not updated within the
last 1 year+. I am therefore setting this issue to 'Verified' as the first step
towards Closing it. If you feel this is incorrect, please re-open the issue and
add any comments.

Many thanks,
Andrew
 
Cleaning-up and Closing old Issues
~ The Grand Bug Squash, pre v3 ~
http://marketing.openoffice.org/3.0/announcementbeta.html
Comment 13 ace_dent 2008-05-17 23:02:16 UTC
As per previous posting: Verified -> Closed.
A Closed Issue is a Happy Issue (TM).

Regards,
Andrew