Re: QA:Testcase_Mediakit_ISO_Size for Netinst
On Thu, Aug 12, 2010 at 09:12:36 +, Andre Robatino wrote: > > CD: 734003200 bytes or 716800 blocks > 1 GB Live: 10 bytes or 976563 blocks > DVD: 47 bytes or 4589844 blocks Note that you probably want to limit images to 4 GiB so that they can be downloaded to FAT file systems. That's a bit of a hit to DVDs, but will probably make up for it in saved aggravation. The Spins SIG current policy is that live images for which we produce ISOs need to be less than 4 GiB. -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: QA:Testcase_Mediakit_ISO_Size for Netinst
I have updated the test case - added example, simplified. What do you think? - "Andre Robatino" wrote: ... https://fedoraproject.org/wiki/QA:Testcase_Mediakit_ISO_Size ... -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: QA:Testcase_Mediakit_ISO_Size for Netinst
Kamil Paral redhat.com> writes: > > - "Andre Robatino" fedoraproject.org> wrote: > > > I removed the specific examples (which also has the advantage of one > > less thing > > to have to update periodically). > > When I look at it know, I must say the examples were quite handy. I would > put there at least one example. Probably right under bullet 1 in How to > test. It simplifies thinking a bit :) Since netinst is now listed as one of the CD-size media, I would have to add it to the first example. But the old examples referred to 13 Beta, and the Alpha and Beta images were recently deleted, so I don't know what the netinst size was. I was unhappy with the examples anyway, due to the possibility that displaying anything other than byte sizes might lead to rounding/truncation errors, and the fact that one had to be careful about when to use or not use the --si option. > > I'd also like to put in exact byte > > size numbers > > for the lower limits as well, for consistency. As I understand it, the > > exact > > values aren't critical. Should it say 2 MB = 2,000,000 bytes, or 2 MiB > > = > > 2,097,152 bytes? > > Depends on the example, but I would maybe stick to the default output > of ls -s, therefore IEC units. If all the numbers are in IEC units then > one example is enough. But most newer media (with the notable exception of CDs, which will be obsolete eventually) uses SI units, and their sizes in bytes are easier to remember than the sizes in 1024-byte blocks. Even when the media uses IEC units, the sizes in blocks aren't necessarily easier to remember. CD: 734003200 bytes or 716800 blocks 1 GB Live: 10 bytes or 976563 blocks DVD: 47 bytes or 4589844 blocks > > Should the upper size limit for all the Live media be 1 GB, or should > > it be > > CD-size for some of them, and 1 GB for the rest? > > AFAIK we have only one live image media, right? And that is supposed > to be sized under 1 GB. I thought the live images were expected to be CD-sized for KDE and 1 GB for Gnome, or something like that. -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: QA:Testcase_Mediakit_ISO_Size for Netinst
- "Andre Robatino" wrote: > I removed the specific examples (which also has the advantage of one > less thing > to have to update periodically). When I look at it know, I must say the examples were quite handy. I would put there at least one example. Probably right under bullet 1 in How to test. It simplifies thinking a bit :) > I'd also like to put in exact byte > size numbers > for the lower limits as well, for consistency. As I understand it, the > exact > values aren't critical. Should it say 2 MB = 2,000,000 bytes, or 2 MiB > = > 2,097,152 bytes? Depends on the example, but I would maybe stick to the default output of ls -s, therefore IEC units. If all the numbers are in IEC units then one example is enough. > > Ideally, I don't think there should be a lower size limit at all, but > a separate > test to detect directly whatever fault would cause such small images. I also think the lower size limit is quite artificial and doesn't have to be specified. If the image is broken and too small, it fail some other test (probably boot test). I would remove lower size limit check completely, another step for simpler test cases. > > Should the upper size limit for all the Live media be 1 GB, or should > it be > CD-size for some of them, and 1 GB for the rest? AFAIK we have only one live image media, right? And that is supposed to be sized under 1 GB. -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: QA:Testcase_Mediakit_ISO_Size for Netinst
I removed the specific examples (which also has the advantage of one less thing to have to update periodically). I'd also like to put in exact byte size numbers for the lower limits as well, for consistency. As I understand it, the exact values aren't critical. Should it say 2 MB = 2,000,000 bytes, or 2 MiB = 2,097,152 bytes? Ideally, I don't think there should be a lower size limit at all, but a separate test to detect directly whatever fault would cause such small images. Should the upper size limit for all the Live media be 1 GB, or should it be CD-size for some of them, and 1 GB for the rest? https://fedoraproject.org/wiki/QA:Testcase_Mediakit_ISO_Size -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: QA:Testcase_Mediakit_ISO_Size for Netinst
On Thu, 2010-08-12 at 02:30 -0400, Kamil Paral wrote: > - "Andre Robatino" wrote: > > > James Laska redhat.com> writes: > > > > > I believe the {netinst,boot}.iso are intended to fit on a 700M cdr. > > > > I edited > > > > https://fedoraproject.org/wiki/QA:Testcase_Mediakit_ISO_Size > > > > to indicate that netinst should be CD-sized. I'd like to delete the > > examples, > > which all use "ls -sh", due to the possibility for confusion between > > IEC values > > (which are displayed by default) and SI values (which require the --si > > option). > > The upper size limits are already given as exact numbers of bytes, so > > people > > can just use "ls -l" (which is easier to remember and not subject to > > rounding/truncation errors). Is this okay? > > I'm not sure which way it is easier to compare it :) But go ahead and try > it and we will see. > > Quite interestingly, 'ls -s --si' (without -h) prints rounded numbers > (like 400M) instead of precise numbers, at least for me. That's really > not ideal. --si is likewise -h, just use powers of 1000 not 1024(see 'ls' manual). -- Contacts Hurry FAS Name: Rhe Timezone: UTC+8 TEL: 86-010-62608141 IRC nick: rhe #fedora-qa #fedora-zh -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: QA:Testcase_Mediakit_ISO_Size for Netinst
- "Andre Robatino" wrote: > James Laska redhat.com> writes: > > > I believe the {netinst,boot}.iso are intended to fit on a 700M cdr. > > I edited > > https://fedoraproject.org/wiki/QA:Testcase_Mediakit_ISO_Size > > to indicate that netinst should be CD-sized. I'd like to delete the > examples, > which all use "ls -sh", due to the possibility for confusion between > IEC values > (which are displayed by default) and SI values (which require the --si > option). > The upper size limits are already given as exact numbers of bytes, so > people > can just use "ls -l" (which is easier to remember and not subject to > rounding/truncation errors). Is this okay? I'm not sure which way it is easier to compare it :) But go ahead and try it and we will see. Quite interestingly, 'ls -s --si' (without -h) prints rounded numbers (like 400M) instead of precise numbers, at least for me. That's really not ideal. -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: QA:Testcase_Mediakit_ISO_Size for Netinst
James Laska redhat.com> writes: > I believe the {netinst,boot}.iso are intended to fit on a 700M cdr. I edited https://fedoraproject.org/wiki/QA:Testcase_Mediakit_ISO_Size to indicate that netinst should be CD-sized. I'd like to delete the examples, which all use "ls -sh", due to the possibility for confusion between IEC values (which are displayed by default) and SI values (which require the --si option). The upper size limits are already given as exact numbers of bytes, so people can just use "ls -l" (which is easier to remember and not subject to rounding/truncation errors). Is this okay? -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: QA:Testcase_Mediakit_ISO_Size for Netinst
On Wed, 2010-08-11 at 14:33 +, Andre Robatino wrote: > My assumption has been that since the content of netinst is basically a subset > of CD #1, it should be CD-sized. If this is correct, it could be made > explicit. I believe the {netinst,boot}.iso are intended to fit on a 700M cdr. Thanks, James signature.asc Description: This is a digitally signed message part -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: QA:Testcase_Mediakit_ISO_Size for Netinst
My assumption has been that since the content of netinst is basically a subset of CD #1, it should be CD-sized. If this is correct, it could be made explicit. -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
QA:Testcase_Mediakit_ISO_Size for Netinst
In our F14 installation matrix [1] we have testcase "QA:Testcase_Mediakit_ISO_Size" specified for "Boot.iso/Netinst.iso installation" section. However that test case does not mention any requirements for boot.iso/ netinst.iso. Do we have some size requirements on these images? What are they? Or should we just leave out this test case from that section? Thanks, Kamil [1] https://fedoraproject.org/wiki/Test_Results:Fedora_14_Alpha_RC3_Install -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test