Re: Added support for file list in single file update

[Shigio YAMAGUCHI]

Hello,

> Hello,

>  I wanted to an ability to update existing tags based on incremental

> changes. My incremental changes are on a small bunch of files. Here is

> the workflow.

>

> 1. Create a p4 client (we use perforce with some cool snapshot based

> client creation at work)

> 2. I get a pre-built p4 client with gtags built

> 3. Edit files and update tags (gtags --single-update file.cpp). Invoke

> the command for each file that I have modified

> 4. Other developers have checked in their code over time

> 5. I create a new p4 client and import my changes into the new client

> 6. Again, I update my local tags

>

> When tags are pre-built during build between snapshots, I know the

> files that are added/modified/deleted. I wanted to improve the

> performance by not calling 'gtags' binary multiple times (once per

> file) for an update AND not use '-i' and let 'gtags' stat() all files

> to find if deleted.

Although I don't know 'perforce' and some terms like 'p4 client' and

'pre-built p4', it seems that your suggestion is changing 'single-update'

into 'multi-update'. If so, I agree with you.

> The following patch implements 2 features:

>

> - "GTAGSLOGICALPATH" for supporting logical path instead of realpath

> - treat the file given to --single-update as a single file or a file

> containing a list of files

> -- It supports optional prefixes in the file list '-' or '+' to avoid

> costly stat() over NAS (I work in storage and know the overheads of

> stat() to try avoiding it)

>

> Sample file list (4 lines):

> -path/to/deleted/file.cpp

> +path/to/new/file.cpp

> +path/to/modified/file.cpp

> path/to/a/file/and/let/stat/determine/if/present/or/missing.cpp

I'm hesitant to accept the prefixes (+, -), because wrong specification

of them corrupts the tag files.

How effective is this?

--

Shigio YAMAGUCHI <address@hidden>

PGP fingerprint: D1CB 0B89 B346 4AB6 5663  C4B6 3CA5 BBB3 57BE DDA3