| 1 |
On 06/12/2016 03:05 AM, Michał Górny wrote: |
| 2 |
> Dnia 12 czerwca 2016 11:49:26 CEST, Zac Medico <zmedico@g.o> napisał(a): |
| 3 |
>> On 06/12/2016 02:28 AM, Michał Górny wrote: |
| 4 |
>>> Dnia 12 czerwca 2016 11:10:55 CEST, Zac Medico <zmedico@g.o> |
| 5 |
>> napisał(a): |
| 6 |
>>>> On 05/22/2016 01:21 AM, Michał Górny wrote: |
| 7 |
>>>>> Introduce a new logic for INSTALL_MASK handling in merging code, |
| 8 |
>>>>> replacing the old code that removed matching files and directories |
| 9 |
>>>>> from imagedir in bash. The new code actually ignores matching files |
| 10 |
>>>>> on-the-fly while testing for file collisions and merging files. |
| 11 |
>>>>> The files are still written to CONTENTS, and output using "###" |
| 12 |
>> zing |
| 13 |
>>>>> to indicate being masked, yet are not actually merged to the |
| 14 |
>>>> filesystem. |
| 15 |
>>>> |
| 16 |
>>>> Since collision-protect relies on existing files in its collision |
| 17 |
>> test, |
| 18 |
>>>> install-masked files are no longer going to trigger collisions. |
| 19 |
>> Then, |
| 20 |
>>>> since the install-masked files are still written to CONTENTS, it's |
| 21 |
>>>> possible for the unmerge of one package to unmerge colliding files |
| 22 |
>> that |
| 23 |
>>>> belong to another package! |
| 24 |
>>>> |
| 25 |
>>>> There are a number of ways to solve this problem. For example, we |
| 26 |
>> could |
| 27 |
>>>> have the unmerge code ignore any files in CONTENTS that match the |
| 28 |
>>>> INSTALL_MASK value that was used at merge time. |
| 29 |
>>> |
| 30 |
>>> Hmm, thinking about this more widely (i.e. actually thinking rather |
| 31 |
>> than mimicking the old behavior), I think it would be better to |
| 32 |
>> actually use the original file set for collision-protect. This will |
| 33 |
>> make it possible to detect collisions that would otherwise be hidden |
| 34 |
>> via INSTALL_MASK. |
| 35 |
>> |
| 36 |
>> Even then, we have to carefully consider how the absence of installed |
| 37 |
>> files affects the collision test. It's safest for the unmerge code to |
| 38 |
>> be |
| 39 |
>> aware of the merge time INSTALL_MASK setting, and not try to unmerge |
| 40 |
>> the |
| 41 |
>> install-masked files. |
| 42 |
> |
| 43 |
> But then it wouldn't unmerge the newly masked files as well. |
| 44 |
|
| 45 |
I'm suggesting to save the INSTALL_MASK setting from merge time, so any |
| 46 |
new INSTALL_MASK settings since then are irrelevant. |
| 47 |
|
| 48 |
> Getting this right will require a lot of effort, and we're less likely to screw something up if we keep it simple. |
| 49 |
|
| 50 |
You can use exactly the same pattern matching code for merge and |
| 51 |
unmerge. It's worthwhile, given the say that collision-protect relies on |
| 52 |
file existence in the collision test. |
| 53 |
-- |
| 54 |
Thanks, |
| 55 |
Zac |
Archives