Hi Rokh, See below.
On 10/09/09 01:33, Rokh Wang wrote: > Hi Sue and Frank, > I really appreciate your careful review. The assertions have been > revised according to your comments. > Would you please give a quick review again. If the assertions documents > is OK. I'll start the coding > work based on it. > //Thanks a lot! > -Rokh > > Susan Sohn wrote: >> On 09/28/09 02:40, Rokh Wang wrote: >>> Susan Sohn wrote: >>>> Hi Rokh, >>>> >>>> Frank and I sat down and went through these assertions together. >>>> Here are our comments. >>>> >>>> General comments: >>>> The text installer will be able to create Extended partitions and >>>> Solaris partitions. The test assertions (e.g., manual_04, >>>> tp_partition_003, and others) assume the ability to create other >>>> kinds of partitions such as pcfs. Assertions that make this >>>> assumption should be revised. For instance, manual_04 could be >>>> revised so that fdisk is used to create the extended/logical >>>> partitions. Also, no partition numbers will be displayed on the text >>>> install screens. Setting a partition to unused should make its size 0. >>>> >>> Hi Sue, >>> I have revised all the assertions based on your comments. For the >>> partition numbers, I just use them to be >>> the identify for partitions. >>> B.T.W. according feedback from Jean, now p1-p4 are primary/extended >>> partitions and p5-p36 are the logical >>> partitions, so I also revised all the assertions using this numbering >>> scheme. >>> >>> Here I have two questions: >>> 1. Two extended partition problem. >>> As I know that >>> ======================================================================= >>> The GUI installer does allow there to be more than one Solaris partition >>> displayed, but it is reported as an error when the user clicks Next. The >>> user can't proceed until there is one and only one Solaris partition. >>> ======================================================================= >>> Since there only one extended partition can be exist on one disk, so >>> I have a question if text installer >>> check if one extended partition exists, user can't create another >>> primary partition as extended partition? >> >> Only one extended partition will be allowed at a time. >> >>> 2. Using F5 delete partition problem >>> Now to delete a partition, I use F5 on the "Select Partition" screen >>> to change a partition type to "unused". >>> Would you please tell me how you implement this. Just change the id >>> of the partition to 0 or a number that >>> represents "unused" or other method? >> >> The partition type is set to UNUSED, which is defined as 0. >> >>> B.T.W. I have added a new slice assertion - tp_slice_008 to test on >>> SPARC when there is no valid vtoc >>> information on the test disk. >>> >>> Would you please take a look at the embedded feedback too. >>> >>> Thanks, >>> -Rokh >>> >>>> manual_04: the user must create a Solaris partition before leaving >>>> the Select Partition screen. >>> I have revised this case. >> >> ok >> >>>> tp_partition_006, once the partitions have been set to unused in >>>> step 4, the size should be 0. >>> I have revised this case. >> >> ok >> >>>> tp_partition_007: step 6, unused size should be 0. >>> I have revised this case. >>>> tp_partition_09 step 6. When you recreate the extended partition, the previous logical partitions will be gone. The available space for partition 1 is uncertain. >>>> tp_partition_010: Per Frank's email yesterday, you will not be able >>>> to cycle back to Extended, so you might not want this test case. >>> I have deleted this case. >>> Now the tp_partition_010 is a new test case to test that how "Text >>> Installer" deal with unrecognized partitions. >>> I created partitions using id 231,232,233,234 as partitions whose >>> type is not recognized. >>> Would you please help me make sure such partitions will be recognized >>> as "Unknow" partitions by text installer. >> >> Recognizing them as "Unknown" is the expected behavior. Step 3, "Unknow" should be "Unknown" >>>> tp_partition_011: Step 5 is invalid. Only Solaris and extended >>>> partitions can be resized. You could delete partition 3 if desired >>> Got it, I revised all the cases that test resizing partitions. In all >>> the Resizing assertions, only the Solaris and extended partitions >>> can be resized. And tp_partition_011 now is designed to test >>> "Resizing primary partition and extended partition". >> >> In step 6, unused #1 size should be 0. > Revised ok >>>> tp_partition_012: Per Frank's email yesterday, you will not be able >>>> to do step 3 and step 8. The partition must be empty to resize. >>> I have deleted this case. >>>> >>>> tp_partition_013: step 5 is invalid. Only Solaris and extended >>>> partitions can be resized. >>> Now tp_partition_012 is designed to test "Increasing logical >>> partition", and tp_partition_013 is designed to test "Decreasing >>> logical partition". >> >> This test case has a couple of issues. First, as said above, only >> Solaris and extended partitions can be resized. You are resizing a >> pcfs partition - you should say you are resizing the Soalris2 >> partition. Also, you started out with a pcfs and Soalris2 partition >> but somehow ended up with two pcfs partitions. Assuming that the >> second pcfs partition (partition 8) in step 4 is a Soalris2 partition, >> and that is the one you are resizing, then it looks ok. >> > Revised ok >>>> tp_partition_014: steps 3-5 aren't valid. Step 6 is allowed. Also, >>>> note that you would need to create a solaris partition in order to >>>> proceed to the next screen. >>> I have revised this case. >> >> In step 1, available space for extended partition should be 70%. >> The avail space for the linus partition should be 50%. >> Step 5 is invalid. You can't have two solaris2 partitions. >> It would also be nice to show the expected display of the partitions >> after steps 3-5. > Revised ok >>>> tp_partition_015: step 3, you can't change pcfs to solaris until >>>> existing solaris partition is deleted. If you switch step 4 to make >>>> solaris partition into unused, and swap steps 3 and 4, that would work. >>> I have revised this case. >> >> ok For tp_partition_015, you should add the expected results of the changes (what you are verifying). >>>> tp_slice_001: Is this test for both x86 and sparc? If so, step 2 >>>> should reference both fdisk partition screen as well as Solaris Slices. >>> Now this case just test sparc. I use tp_partition_002 cover such test >>> on x86. >> >> for tp_slice_001, you should verify that the entire disk is used. >> >> for tp_partition_002, if you select to use the whole disk on the fdisk >> screen, you will never see the screen in step 3. >> > Revised >>>> tp_slice_002: If you destroy the existing vtoc info, then in x86, >>>> you will not see the slice screens. For the sparc case, you can't >>>> create "default" slices. >>> I have revised this case. Would you please help me review if the >>> assertion is OK. >> >> The only thing we can create on slice screens are the zfs root pool. >> Existing slices can be preserved, but that is all that you can create. > Revised ok >>>> tp_slice_003: There are two step 4's. For the second step 4, the >>>> size of slice 0 should be 0. >>>> >>>> tp_slice_004: After much discussion, we believe the outcome should be: >>>> slice 0: size=0 avail=10 >>>> slice 1: size=15, avail=100 >>>> slice 3 size=0, avail = 75 >>> I have revised this case. >> >> ok >> >>>> tp_slice_005: Step 4 - you can't create a second ZFS Root pool if >>>> one already exists. Couldn't really review the rest of this case. >>>> >>>> tp_slice_005: Step 4 should reference Select Slice screen, similar >>>> to step 3. >>> I have revised this case. >> >> Step 5 - which unused are you changing? It's hard to tell what is >> going on in this case. Step 6 - you can't cycle back to the original >> type. You can use F7 to completely reset. >> > Revised You should add the expected results of the changes (what you are verifying). The changes now appear to be minor, so no need for further review. Thanks, Sue and Frank
